Navigating autonomous coverage robots

Abstract

A method of navigating an autonomous coverage robot between bounded areas includes positioning a navigation beacon in a gateway between adjoining first and second bounded areas. The beacon configured to transmit a gateway marking emission across the gateway. In some example, the navigation beacon may also transmit a proximity emission laterally about the beacon, where the robot avoids cleaning the migration within the proximity emission. The method also includes placing the coverage robot within the first bounded area. The robot autonomously traverses the first bounded area in a cleaning mode and upon encountering the gateway marking emission in the gateway, the robot remains in the first bounded area, thereby avoiding the robot migration into the second area. Upon termination of the cleaning mode in the first area, the robot autonomously initiates a migration mode to move through the gateway, past the beacon, into the second bounded area.

Description

TECHNICAL FIELD

This disclosure relates to autonomous coverage robots and associated navigation systems.

BACKGROUND

Autonomous robots are robots which can perform desired tasks in unstructured environments without continuous human guidance. Many kinds of robots are autonomous to some degree. Different robots can be autonomous in different ways. An autonomous coverage robot traverses a work surface without continuous human guidance to perform one or more tasks. In the field of home, office and/or consumer-oriented robotics, mobile robots that perform household functions such as vacuum cleaning, floor washing, patrolling, lawn cutting and other such tasks have been widely adopted.

SUMMARY

According to one aspect of the invention, a robot features a chassis, a wheeled drive supporting the chassis and operable to maneuver the robot, sensors responsive to an environment about the robot, a memory and a controller. The memory stores instructions configured as a set of behaviors to cause the wheeled drive to move the robot according to a predetermined sequence of the behaviors of the set, at least one of the behaviors of the set featuring a sensor-based enabling condition dependent upon a sensor detection and at least one of the behaviors of the set featuring a sequence-based enabling condition dependent upon a completion of another behavior in the set. The memory also stores instructions configured as an arbiter to enable control of the wheeled drive according to a behavior priority, the arbiter passing on commands from a highest priority behavior to control the wheeled drive when enabling conditions of the highest priority are satisfied. The controller is in communication with the wheeled drive, the sensors, and the memory, and executes the arbiter to control the wheeled drive according toe the highest priority behavior. The behaviors of the set tend to control the wheeled drive in the predetermined sequence as the sensor-based enabling conditions and the sequence-based enabling conditions are satisfied.

In some embodiments, the set of behaviors has a sequence-based enabling condition from a parent set of behaviors, the set of behaviors being executed in the predetermined sequence upon enablement and a predetermined sequence of the parent set of behaviors being resumed upon either completion of the set of behaviors or cessation of the enabling condition.

In some cases, the set of behaviors has a sensor-based enabling condition, the controller discontinuing execution of the predetermined sequence upon cessation of the enabling condition.

In some applications, behaviors of the set having a sequence-based enabling condition have a lower priority than behaviors of the set having only sensor-based enabling conditions. For example, behaviors of the set having a sequence-based enabling condition may have a lower priority than at least an escape behavior and an avoidance behavior.

In some configurations, a docking behavior having a sensor-based enabling condition has a higher priority than both a navigation behavior having a sensor-based enabling condition and a sequence-based enabling condition. In some configurations, a coverage behavior having a sensor-based enabling condition has a lower priority than both a navigation behavior having a sensor-based enabling condition and a sequence-based enabling condition.

In some embodiments the controller aborts execution of a behavior upon cessation of the associated enabling condition.

At least one behavior of the set may be a sequenced-reactive behavior having at least one sensor-based enabling condition and at least one sequence-based enabling condition. At least partial execution of another behavior of the set may enable the sequence-based enabling condition of the sequenced-reactive behavior, and the controller may abort the sequenced-reactive behavior upon cessation of at least one of the enabling conditions of the sequence-reactive behavior. In some examples at least two sequenced-reactive behaviors of the set share at least one sensor-based enabling condition or sequence-based enabling condition, thereby allowing multiple potential entry points into the predetermined sequence upon satisfying the shared enabling condition.

The sensor-based enabling condition may include detecting a navigation pose, or detecting an obstacle proximate the robot, for example.

In some cases the memory also stores a sequence planner executable by the controller to monitor a current state of the predetermined sequence and to set a sequence-based enabling condition according to a current sequence state of the robot upon detection of a sequence initiating event. The sequence planner may be configured to set a sequence-based enabling condition according to a current location state of a monitored topological state map, for example. The sequence planner may be configured to instruct the arbiter to allow execution of a behavior by the controller based on a monitored state of a finite state machine stored in the memory. In some arrangements the sequence planner controls execution of the controller of at least one goal oriented sequence of behaviors, the behaviors of the goal oriented sequence having a sequence-based enabling condition set by the sequence planner according to a current sequence state of the robot upon detection of a goal oriented sequence initiating event.

Another aspect of the invention features a method of navigating a robot, the method including monitoring whether sensor-based enabling conditions of behaviors are satisfied, each sensor-based enabling condition being dependent upon a sensor detection; monitoring whether sequenced-based enabling conditions of behaviors are satisfied, each sequenced-based enabling condition being dependent upon a successful execution of a preceding behavior; arbitrating among the behaviors to permit control by the behaviors according to a priority, by permitting control by a highest priority behavior when the enabling conditions of the highest priority behavior are satisfied; and driving the robot to end to navigate according to a predetermined sequence of control by the behaviors as the sensor-based enabling conditions and the sequence-based enabling conditions of the behaviors are successively satisfied.

In some embodiments arbitrating among the behaviors includes arbitrating among a set of behaviors to be executed in a predetermined sequence, the set having a sequence-based enabling condition from a parent set of behaviors; executing the set of behaviors in the predetermined sequence upon enablement, and resuming a predetermined sequence of the parent set of behaviors upon completion of the set of behaviors or cessation of the enabling condition.

In some cases, arbitrating among the behaviors includes arbitrating among a set of behaviors to be executed in a predetermined sequence, the set having a sensor-based enabling condition; and discontinuing execution of the predetermined sequence upon cessation of the enabling condition.

Behaviors having a sequence-based enabling condition may have a lower priority than behaviors of the set having only sensor-based enabling conditions. Behaviors having a sequence-based enabling condition may have a lower priority than at least an escape behavior and an avoidance behavior.

In some examples, a docking behavior having a sensor-based enabling condition has a higher priority than both a navigation behavior having a sensor-based enabling condition and a sequence-based enabling condition.

In some examples, a coverage behavior having a sensor-based enabling condition has a lower priority than both a navigation behavior having a sensor-based enabling condition and a sequence-based enabling condition.

Some examples of the method include aborting execution of a behavior upon cessation of the associated enabling condition.

In some embodiments, at least one behavior includes a sequenced-reactive behavior having at least one sensor-based enabling condition and at least one sequence-based enabling condition. At least partial execution of another behavior of the set enables the sequence-based enabling condition of the sequenced-reactive behavior, and the robot aborts the sequenced-reactive behavior upon cessation of at least one of the enabling conditions of the sequenced-reactive behavior. In some cases at least two sequenced-reactive behaviors share at least one sensor-based enabling condition or sequence-based enabling condition, thereby allowing multiple potential entry points into the predetermined sequence upon satisfying the shared enabling condition.

In some cases the sensor-based enabling condition includes detecting a navigation pose or detecting an obstacle proximate the robot.

Some examples also include monitoring a current state of a predetermined sequence of behaviors, and setting a sequence-based enabling condition according to a current sequence state of the robot upon detection of a sequence initiating event. The sequence-based enabling condition may be set according to a current location state of a monitored topological state map, for example. The step of arbitrating may include executing at least one goal oriented sequence of behaviors, the behaviors of the goal oriented sequence having a sequence-based enabling condition set by the sequence planner according to a current sequence state of the robot upon detection of a goal oriented sequence initiating event.

DESCRIPTION OF DRAWINGS

FIG. 1A is a perspective view showing an example of an autonomous coverage robot.

FIG. 1B is an exploded views of an autonomous coverage robot.

FIG. 1C is an exploded views of a bumper of an autonomous coverage robot.

FIG. 2 illustrates the positions of omni-directional receiver and directional receiver on the bumper of an autonomous coverage robot.

FIG. 3 is a perspective view of a directional receiver.

FIG. 4 is a front view of a directional receiver.

FIG. 5 is an exploded view of the directional receiver.

FIGS. 6-8 show perspective, side, and cut-away views of the omni-directional receiver.

FIG. 9 shows a below-perspective view autonomous coverage robot.

FIG. 10 shows an exploded view of an example navigation beacon.

FIGS. 11 and 12 show perspective and front views of an example beacon emitter assembly.

FIG. 13 shows an example of a simplified navigation beacon that may be used with an autonomous mobile robot navigation system.

FIGS. 14-17 slow various views of an example base station that may be included in the autonomous mobile robot system.

FIGS. 18-21 shows perspective, top, rear, and side views of an example emitter assembly used in base station.

FIG. 22 shows a block diagram of representative electronic structure of an autonomous coverage robot.

FIG. 23A shows a first example of applying of a method of navigating an autonomous coverage robot between bounded areas using navigation beacons with the addition of a base station; and FIG. 23B shows a second example in which the robot encounters more than one beam at the same time.

FIGS. 24A and 24B are block diagrams illustrating a software architecture for navigating an autonomous coverage robot.

FIGS. 25A-C show schematic views of a robot moving from one area to another using a navigation beacon.

FIG. 26 shows a schematic view of a beam align behavior of a robot.

FIG. 27 shows a schematic view of a directed beam homing behavior of a robot.

FIG. 28 shows a schematic view of a field align behavior of a robot.

FIG. 29 shows a schematic view of a field following behavior of a robot.

FIG. 30 shows a schematic view of a leaving beacon behavior of a robot.

FIGS. 31A-H show schematic views of a robot moving from one area to another using a navigation beacon.

FIGS. 32A-E show schematic views of a robot docking with a base station.

Like reference symbols in the various drawings indicate like elements.

DETAILED DESCRIPTION

FIGS. 1A and 1B show above-perspective and exploded views of an autonomous coverage robot 100. The robot 100 has a chassis 102, a controller 103, an omni-directional receiver 104, and a directional receiver 106. Chassis 102 has a forward drive-direction and carries controller 103 and the receivers 104 and 106 on a bumper 107. Receivers 104 and 106 provide navigation information to controller 103. Using input from receivers 104 and 106, controller 103 generates commands to be carried out by the robot 100. As a result, the robot 100 is capable of cleaning floors in an autonomous fashion.

FIG. 2 illustrates the positions of omni-directional receiver 104 and directional receiver 106 on the bumper 107 of the robot 100. Bumper 107 may also have other sensors, as shown in FIG. 1C, to aid the robot 100 as it navigates about its surroundings. Proximity sensors 1072 may be used to determine when an obstacle is close to the robot 100. Contact sensors 1074 may be used to determine when the robot 100 has physically encountered an object. Cliff sensors 1076 may be used to sense when the robot 100 has encountered the edge of the floor, such as when it encounters a set of stairs.

Omni directional sensors 104 may be used to sense when the robot 100 is in close proximity to a navigation beacon. For example, the omni-directional sensor 104 may relay a signal to a control system that indicates the strength of an emission, where a stronger signal indicates closer proximity to a navigation beacon.

Proximity sensors 1072 may be used to detect when an obstacle is nearby. The proximity sensors 1072 may, for example, be infrared or ultrasonic sensors that provide a signal when an object is within a given range of the robot 100.

Contact or bump sensors 1074 may be used to detect if the robot 100 physically encounters an obstacle. Such sensors 1074 may use a physical property such as capacitance or physical displacement within the robot 100 to determine when it has encountered an obstacle.

Cliff sensors 1074 may be used to sense when the robot 100 has encountered the edge of the floor, such as when it encounters a set of stairs. The robot 100 may have behaviors that cause it to take an action, such as changing its direction of travel, when an edge is detected.

In some implementations, cliff sensors 1074 may be installed within a mounting apparatus that stabilizes and protects the sensor and which positions the sensor to point towards the window installed onto the bottom of the mounting apparatus. Together the sensor, the mounting apparatus and the window comprise a cliff sensor unit. There may, for instance, be four cliff sensor units installed within the bumper.

Reliability of a cliff sensor may be increased by reducing dust buildup. In some implementations, a window may be installed on the bottom of the mounting apparatus which includes a shield mounted within a slanted molding composed of a material which prevents dust build up, such as an antistatic material. The shield component and the molding may be welded together. To further facilitate the reduction in dust and dirt buildup, the shield may be mounted on a slant to allow dirt to more easily slide off.

In some implementations, a secondary cliff sensor may be present behind existing cliff sensors to detect floor edges in the event that a wheel drop sensor on the caster wheel fail.

Controller 103 may be configured to propel the robot 100 according to a heading setting and a speed setting. Signals received from proximity and contract sensors may be used by the control system to issue commands that deal with obstacles. Signals from the proximity and contact sensors may cause the control system to change the commanded speed or heading of the robot 100. For instance, a signal from a proximity sensor due to a nearby wall may result in the control system issuing a command to slow down. In another instance, a collision signal from the contact sensors due to an encounter with a chair may cause the control system to issue a command to change heading. In other instances, the speed setting of the mobile the robot may not be reduced in response to the contact sensor; and/or the heading setting of the mobile the robot may not be altered in response to the proximity sensor, but instead the two sensors may operate independently.

Controller 103 may include a first independent behavioral routine configured to adjust the speed setting of the mobile the robot; and a second independent behavioral routine configured to alter the heading setting of the mobile the robot, in which the first and second independent behavioral routines are configured to execute concurrently and mutually independently. The first independent behavioral routine may be configured to poll the proximity sensor, and the second independent behavioral routine may be configured to poll the kinetic bump sensor.

FIG. 3-5 show perspective, front, and exploded views of the directional receiver 104. The directional receiver 104 is positioned on the front of the chassis 102. Emissions may be received along the drive direction by the directional receiver 104 which generates and sends corresponding signals to controller 103. The directional receiver 104 includes a mounting receptacle 108 with a first aperture 110 and a second aperture 112 that are aligned with the drive direction of the chassis 102. Associated with the apertures 110, 112 are a first component receiver 114 and a second component receiver 116.

The component receivers 114, 116 are positioned relative to the apertures 110, 112 such that emissions which originate along the driving direction and fall on the directional receiver 104 may pass through the apertures 110, 112 and onto their respective component receivers 114, 116. Emissions falling on the component receivers 114, 116 result in a corresponding signals being generated which may be used by controller 103 to determine the direction of the emission relative to the robot's drive direction. A cover 117 attached to the top of the mounting receptacle 108 to prevent emissions that do not originate along the driving direction from falling on the component receivers 114, 116. In some implementations, locking tabs 119 or some other fastening method may be used to secure the cover 117 to the mounting receptacle 108.

In some implementations, the component receivers 114, 116 of the directional receiver 104 can detect a homing beam with an overlap of substantially between 10% and 30%. In one example, the first and second component receivers 114, 116 may be positioned such that emissions falling squarely on each would form an angle in the range of 1 to 15 degrees. In another example, the first and second component receivers 114, 116 may be aligned so emissions aligned with them cross at a point 3 to 5 meters in front of the robot. In one example, each component receiver 114, 116 receives emissions within a sweep angle of between about 28 to 33 degrees with a center of the sweep at about 14 degrees from a direction normal to the directional receiver 104 and a sweep overlap of about 10 degrees with the other component receiver 114, 116.

In some cases, the controller 103 may maneuver the robot 100 to detect an emission with the omni-directional receiver 104. The direction of the emission may be determined using the component receivers 114, 116. In response to the detection, the controller 103 may orient the chassis 102 to align the robot drive direction with the determined emission direction.

In other cases, controller 103 may orient the chassis 102 to align the robot drive direction with the determined emission direction based on the signals generated by the component receivers 114, 116 in response to emissions detected with the directional receiver 106.

FIGS. 6-8 show perspective, side, and cut-away views of the omni-directional receiver 104. The omni-directional receiver 104 may include a housing 118, a conical reflector 120 and an emission receiver 122. The housing 118 has an upper portion 124 and an inner cavity 126. The upper portion 124 may allow a transmission of en emission into inner cavity 126. Conical reflector 120 is located on an upper surface of the cavity 126 to reflect emissions falling on the upper portion 124 of the housing 118 into the inner cavity 126. Emission receiver 122 is located in inner cavity 126 below conical reflector 120.

In some implementations, the receivers 114, 116, and 122 may be configured to receive transmissions of infrared light (IR). In such cases, a guide 128 (e.g. a light pipe) may guide emissions reflected off the conical reflector 120 and channels them to emission receiver 122.

FIG. 9 shows a below-perspective view of the autonomous coverage robot 100. A drive system 130 includes a first drive wheel 132 and a second drive wheel 134 which support chassis 102. A castor 136 may provide additional support to chassis 102. Motors may be mechanically coupled to the drive wheels to propel the robot 100, providing forward, reverse, and turning capabilities.

The robot 100 may have a floor cleaning system that removes dirt and debris from the floor as it maneuvers about. The floor cleaning system may include a floor cleaning assembly 140, a cleaning bin 142 and a bin sensor (not shown). The floor cleaning assemblies 140 and 146 and cleaning bin 142 may be carried by the chassis 102. Cleaning bin 142 may be arranged to collect debris removed from a floor being cleaned by cleaning assembly 140. The bin sensor may be configured to generated a signal that indicates whether the cleaning bin has been filled to a threshold. When the threshold has been reached, controller 103 may initiate a docking sequence with a base station (described below). Cleaning bin 142 may be accessed to empty its contents either manually or, in some implementations, the robot 100 may automatically empty cleaning bin 142 when docked.

In addition to providing a framework for the other elements that make up the robot 100, the chassis 102 may have a ridge on its front which is higher than all other points on the chassis 102. Such a ridge may be stop the robot 100 if it encounters an overhead obstacle, such as a couch, and prevent it from becoming wedged underneath. To enable the robot 100 to more easily free itself in the event it becomes wedged, controller 103 may normally command the drive system 130 to operate at less than maximum torque, for instance at 50% of capacity. If the robot 100 becomes a wedged, for instance, sensed by increased current flowing to the drive motors, controller 103 may command increased torque to free the robot 100.

Another example of an anti-wedging system includes a spring loaded wheel system having a potentiometer to measure how much the robot 100 is pushed down by a potential wedging obstacle. Another example includes an infrared sensor disposed on a lower surface of the robot 100, where the infrared sensor is used to measure a distance that the robot 100 is pushed down.

FIG. 10 shows an exploded view of an example navigation beacon 150. Navigation beacon 150 may be used in conjunction with existing walls or other obstacles to create a bounded area. Bounding an area may be done, for example, to restrict a robot from entering or leaving an area. The navigation beacon 150, along with the robot 100 described above, form an example of an autonomous mobile robot system.

Navigation beacon 150 has a beacon emitter assembly 152, which includes a gateway beacon emitter 154 and an omni-directional emitter 160. Gateway beacon emitter 154 may be arranged to transmit a gateway marking emission. For instance, beacon 150 may be located within a gateway that separates a first and second adjacent areas and emit the gateway marking emission to form a boundary. Omni-directional receiver 104 and directional receiver 106 on the robot 100 may detect the gateway marking emissions and thereby function as beacon emission sensors. For example, beacon 150 and the robot 100 may use infrared light (IR) emitters and sensors to create and detect the gateway marking emissions.

In one example, the robot 100 controls the state of the navigation beams through commands transmitted over a packet radio network. The address that the beacons respond to on this network is a combination of a robot address and a node address. After installing batteries in the beacon 150, the beacon 150 periodically tries to contact any robot to see if it should wake up and operate its emitters. A robot 100 may respond by transmitting a radio packet containing an invitation to join its network and a temporary address to sue. While operating with a temporary address, the beacon 150 transmits an infrared code in a fence beam from emitter 154 and force field beam from omni-directional emitter 160 indicating that it is not bound, i.e. its radio address is temporary. When a robot 100 sees a beam with the bound code, it iteratively transmits a radio packet to every recently assigned temporary address to send a new code in its beam called a wink. If the robot 100 sees a wink code, it transmits a radio packet containing a new node address to be used from now on as well as a time indicating the number of hours the address is valid for use. Having bound to a robot 100 successfully, the beacon 150 will henceforth only wake up in the presence of that robot 100. The technique of confirming radio communications using infrared light signals is designed to prevent a robot 100 that is not on the same floor as the beacon 150 from controlling it permanently.

Drive system 130 may be configured to maneuver the robot 100 about the first area while in a cleaning mode. In the cleaning mode, the robot 100 may be redirected in response to detecting the gateway marking emission. In addition, the drive system 130 may also be configured to maneuver the robot 100 through the gateway into the second bounded area while in a migration mode.

While in the cleaning mode, the robot 100 may remain in the first bounded area by changing its drive direction when it detects the gateway marking emission. The gateway marking emission thus acts as a virtual barrier which may prevent the robot 100 from leaving the first bounded area.

In some instances, the drive system 130 may be configured to maneuver the robot 100 about the first bounded area in the cleaning mode for a preset time interval. When the preset time interval elapses, the drive system 130 may be automatically put in migration mode. While in migration mode, drive system 130 may be configured to maneuver the robot 100 across the gateway emission. In other instances, the drive system may configured to maneuver the robot about the first bounded area in the cleaning mode until it encounters of the gateway marking emission a preset number of times. Once the gateway marking emission has been encountered the preset number of times, the migration mode may be automatically initiated.

The robot 100 may include a transmitter for communicating with beacon 150. The transmitter may be used to signals beacon 150 to halt or pause transmission of the gateway emission in the migration mode. By signaling beacon 150 to turn on its various emitters only when needed, the system may implement a power-saving function. Such a function may serve to extend battery life in beacon 150.

FIGS. 11 and 12 show perspective and front views of beacon emitter assembly 152. Beacon emitter assembly 152 includes a first directed vectoring beacon emitter 156 and a second directed vectoring beacon emitter 158. Directed vectoring beam emitters may be used to create an emission field with a characteristic edge and spread pattern that may be used to define a navigation route. Navigation beacon 150 may be located within a gateway between two bounded areas with vectoring beacon emitter 156 arranged to transmit a directed vectoring emission into the first bounded area. The angle between the directed vectoring emission and the gateway may be, for example, in the range of 45-90 degrees. In some instances, the directed vectoring emission may consist of infrared light.

While in migration mode, drive system 130 may be configured to direct the robot 100 toward beacon 150 when it encounters the directed vectoring emission emanating from vectoring beacon emitter 156. The robot 100 may then determine the position of the beacon 150 based on the detected direction of the directed vectoring emission relative to directional receiver 106. Once position is determined, the robot 100 may align itself in a drive direction relative to the directed vectoring emission. For example, the robot 100 may advance forward along the path of the directed vectoring emission to reach and traverse the gateway in which beacon 150 is located.

The robot 100 may be able to remotely activate and deactivate the beacon emissions. For example, the robot 100 may use wireless communication, such as radiofrequency (RF) communication, to pass activation and deactivation signals. The robot 100 may remotely activate the directed vectoring emission of the beacon 150 and deactivate the gateway emission upon initiating the migration mode. The robot 100 may remotely deactivate the directed vectoring emission of beacon 150 and activate the gateway emission upon terminating the migration mode.

In some instances, the beacon 150 may be configured to communicate schedule information with the robot 100. For example, beacon 150 may be able to transmit times for which the robot 100 should enter cleaning mode, when it should enter migration mode, etc. Schedule information may include details such as starting or ending times and dates.

Navigation beacon 150 may also transmit a proximity emission laterally about itself. The robot 100 may avoid cleaning and migration within the proximity emission by executing a pre-configured behavior such as changing its course when the proximity emission is detected. The proximity emission may be thought of as a “force field” through which the robot 100 is not permitted to pass.

Navigation beacon 150 may have a switch to select between a confinement mode, a navigation mode, and an off mode. Navigation beacon 150 may have a range select switch like a virtual wall. There may be a light to indicate when the navigation beacon is operating. The navigation beacon 150 may turn on an operational indicator light it is enabled or only when commanded to do so by the robot 100. There may also be a low battery warning, and there may be a separate low battery light.

In order to reduce power consumption and thus increase battery life, different operational modes may exist for the navigation beacon. When the robot 100 is not operating, navigation beacon 150 may be in a lower power mode in which emitters are off and the navigation beacon periodically monitors the communication link to determine if a wakeup is necessary. When the robot 100 is operating, it may send a signal to a single or multiple navigation beacons to turn on their respective emitters. There may be different commands to turn on the various emitters. In addition, the navigation beacon may return to low power mode after an elapsed time.

In the event multiple navigation beacon are used with the robot 100, each navigation beacon may have a unique identifier, such as a 16-bit (or larger) identification number contained in memory. This identifier may be generated in the factory or by the navigation beacon itself. If the identifier is generated in the factory it may be stored in non-volatile memory in the beacon. The robot 100 may use the unique identifier to generate an internal map which may be used to navigate from one bounded region to another by navigating from one navigation beacon to another. In some implementations the identifier for navigation beacon 150 may be used to generate signals encoded in its emissions that may be detected by the robot 100. In the event that navigation beacon 150 is unable to transmit the entire identification number, a unique code maybe generated as a derivation of the ID or by negotiation between the navigation beacons that are present and/or the robot 100.

The robot and navigation beacons may communicate via point to point or via broadcast transmissions. In the point to point scheme, the robot would have to learn the identifiers of all of the navigation beacons. Then it could send transmissions with a data field indicating the destination navigation beacon. That navigation beacon and only that navigation beacon would respond accordingly. This has the advantage that only specific navigation beacons would respond, keeping other navigation beacons off and thus increasing their battery life. User interaction may be used to train the robot to know the navigation beacons.

An alternate concept is for the robot to broadcast to all navigation beacons a command to enable them. This scheme works without user interaction. In order to mitigate this effect of increased batty consumption and reduced battery life on these navigation beacons, al alleviating technique is to reduce the power of the robot transmission.

FIG. 13 shows an example of a simplified navigation beacon 152 that may be used with an autonomous mobile robot navigation system. In this example, the beacon has a beacon emitter assembly 152 including a gateway emitter 154 and an omni-directional emitter 160, but does not have vectoring emitters. Such a beacon may be used to define boundaries of a region but not necessarily support migration functionality.

FIGS. 14-17 show various views of an example base station 180 that may be included in the autonomous mobile robot system. Base station 180 may be located in a bounded area served by the system. Base station 180 has a base 182 with a defined docking direction and a robot charger 184. The robot 100 may approach base 182 from the docking direction to dock with the base 182. The robot charger 184 is housed in the base 182 and may be used to charge the robot 100 when it is properly docked.

The robot charger 184 may begins operating when triggered by a detection circuit after the robot 100 is docked to base 182. Battery conditions may be used to govern whether a deeply discharge mode, a pre-charge trickle mode, or a post charge trickle mode is used to charge the battery.

FIGS. 18-21 show perspective, top, rear, and side views of an example emitter assembly 186 used in base station 180. Emitter assembly 186 has a lower housing 188 which holds an omni-directional beam emitter 190 and two navigational field emitters 192 and 194.

The omni-directional emitter 190 may be mounted on housing 188 and be configured to project a proximity beam laterally around the docking station through the use of an upper housing 196. Upper housing 196 may be configured such that emissions from omni-directional emitter 19 form a desired pattern outside base 182, such as a circle.

Navigational field emitters 192 and 194 may be attached to lower housing 188 and arranged to emit respective, laterally bounded and overlapping fields of emissions of signal beams. One of the emitted fields may define a lateral field edge aligned with the docking direction and be overlapped by the other emitted fields. The two navigational field emitters of the base station may be positioned such that their angle of overlap may be about 6 degrees and each emitter's aperture angle may be 20-30 degrees. A projection 198 on lower housing 188 may be used to shape the emissions from emitters 192, 194 to achieve the above pattern. In one example, the navigational field emitters 192 projects a beam at about 12 degrees from a direction normal to the emitter assembly 186, where the beam sweeps about 40 degrees starting from about −5 degrees to about 35. The other navigational field emitters 194, with the L-shaped baffle or mask 198 terminating near the middle of the LED, projects a beam at about 12 degrees from a direction normal to the emitter assembly 186, where the beam sweeps from straight out to about −35 degrees. In some implementations, the emission fields may consist of infrared light (IR). In such cases, field emitters 190, 192, and 194 may consist of infrared light emitting diodes (LEDs).

FIG. 22 shows a schematic view of electronics 101 of the robot 100. The electronics 101 include the controller 103 which communicates with a bumper microcontroller 107A that controls the omni-directional receiver 104, directional receiver 106, wall proximity sensors 1072, and bumper switch 1074. The controller 103 monitors all other sensor inputs, including the cliff sensors 1076.

FIGS. 23A-B show an example of applying of a method of navigating an autonomous coverage robot between bounded areas using navigation beacons. The method may be applied to a system such as the example the robot navigation system described above. While the following description relates a particular software architecture and set of algorithms to the navigation beacons disclosed as a physical infrastructure, the very same embodiments of a robot, control and software architecture, and/or many particular algorithms can be used with a different physical infrastructure.

Accordingly, because the manner and methods by which the robot handles sequences of goals and navigation are relatively independent of the actual physical form of the infrastructure, for discussions involving such manner and methods, “navigation beacon” includes active and passive landmarks of other physical kinds (e.g., bar codes, retro-reflective tags, illuminated or IR spots projected on the ceiling or walls, directional RF or visible light, and other features identifiable in the environment), and “beam” includes directed vectors of other physical kinds (including inferred directional vectors calculated or kept in memory that may be followed, as well as line-of-sight directions to distant observed objects). In the example shown, walls 201, a first navigation beacon 202 and a second navigation beacon 203 (also discussed herein as “lighthouses”) define a first bounded area 204, a second bounded area 206, and a third bounded area 207 (which are typically rooms of a home). Navigation beacon 202 may be configured to transmit a gateway marking emission 208 (also noted herein as a “virtual gateway” or “gateway beam”) across a gateway 210 (typically a doorway, but openings of a few feet to 10 or 12 feet are possible, as well as arbitrary segmenting of a room without any particular openings). The gateway marking emission 208 may be an IR signal, for example, which can be sensed by a coverage robot 212.

The robot 212 may be placed within a bounded area 204. The robot 212 may autonomously traverse the bounded area 204 in a cleaning mode. If the robot 212 encounters gateway marking emission 208 in gateway 210, the robot 212 remains in the bounded area 204 without moving into an adjacent bounded area 206. Upon termination of the cleaning or coverage mode in the first area, the robot 212 may autonomously (i.e., in response to an internal trigger condition as discussed herein, e.g., time expired, distance covered, percent covered) enter a migration mode in which the robot continues cleaning but monitors for the gateway 210, locates a gateway 210, moves to and through gateway 210, (past beacon 202,) into the adjacent bounded area 206.

In some implementations, the robot 212 may autonomously enter the cleaning mode after the migration to bounded area 206 has been completed. While in the migration mode, the robot 202 may be non-responsive to gateway emission 208 (although the robot may record having crossed the gateway emission 208). In some cases, the robot 212 may remotely (i.e., by direct or relayed RF communications or line-of-sight or reflected optical signaling) deactivate gateway marking emission 208 upon entering migration mode. The robot 212 may also activate gateway marking emission 208 after exiting the migration mode.

Furthermore, navigation beacon 202 may transmit a proximity emission or field 214 laterally around (surrounding) itself, establishing an exclusion zone or “force field”. The robot 212 may detect this proximity field and use the detection to attempt to avoid maneuvering within the exclusion zone around a beacon 202 while in the cleaning or migrating modes. The primary purpose of the exclusion zone is to prevent the robot 202 from striking the beacon 202 itself and moving it.

Navigation beacon 202 may be configured to transmit a first directed vectoring emission 216 (also discussed herein as a “directed emission,” “directed beam,” “launch beam” or “tractor beam”) into bounded area 204 and a second directed vectoring emission 217 into bounded area 206, with beacon 202 positioned within gateway 210. For example, the robot 212 may drive toward beacon 202 upon detecting directed emission 216 while in the migration mode.

While in the migration mode, the robot 212 may detect the directed emission 216 by sensing the directed emission 216 with a directional receiver 218 (also discussed herein as a “binocular sensor,” e.g., including twin detectors in collimated tubes) located on the front of the robot 212 (i.e. the side of the robot 212 that is aligned with the forward the robot drive direction). Alternatively, the robot 212 may sense the directed emission 216 with an omni-directional receiver 222 prior to detection by directional receiver 218. The term “omni-directional receiver” as discussed herein includes non-directional and multi-directional receivers, e.g., receivers including two or more sensors (at compass points, etc.) directed about the periphery of the robot. The robot 212 may use the signal detected by omni-directional receiver 222 to then maneuver (turn in place or turn going in forward or reverse directions) to sense directed emission 216 with directional receiver 218. The robot 212 may align its direction of travel with a path defined by an edge 220 of directed emission 216. Directional and omni-directional receivers 218 and 222 may have construction and function similar to that described above.

In some cases, the robot 212 may move towards the beacon 202 along the emission path (i.e., serving using the two beam detectors 114 and 116 of the binocular sensor 106) until it sees a beacon perimeter 224 (i.e. the edge of proximity field emission 214). The robot 212 may move along beam perimeter 224 through the gateway 210 and into bounded area 206. The angle between the gateway emission 208 and directed emission 216 may, in some instances, be between about 45 and 90 degrees, and optionally is about 60 degrees. The gateway emission 208 is sized according to the robot diameter, and diverges to be approximately the width of the robot at the proximity field edge or range of detection (e.g., to prevent the robot crossing at that point). This may be a span of 10 degrees or more, but is determined by the robot diameter. The angle between the center of the gateway emission 208 and the center of each directed emission 216 is, in two different examples, about 45 degrees or about 60 degrees, and each directed emission 216 beam is a diverging beam made diffuse by the slot mask near the emitter, and is, in two different examples, about 20-30 degrees (e.g., 25 degrees) or about 30-50 degrees (e.g., 40 degrees). In some instances, the directed emissions 216 and 217 may consist of infrared light.

The robot 212 may remotely activate and/or deactivate emissions coming from beacon 202, such as directed emission 216 or gateway emission 208. Various methods may be used to transmit signals between the robot 212 and the navigation beacons 202 and 203.

In particular, it is effective to use the same inexpensive, common IR-LED emitters for each of the emitters of each of the beacons 202, 203, and in addition for a recharging or other dock 240 for the robot 212. The same sensors on the robot 212 can detect all of the emitters; and different sensors on the robot 212 (e.g., the multi-directional as well as the collimated directional sensors) can detect the same emitters for different purposes (as discussed herein, e.g., following, homing, stopping). In order to differentiate between the different emitters, each emitter may be encoded (e.g., modulated with a different serial code). This also helps avoid confusion with household and other remote controls as well as IR components of sunlight and other ambient sources. However, as shown in FIG. 23, the robot 212 can encounter situations in which it is traveling through the emission paths of several different emitters.

In this situation, especially when using optical multi-directional receivers or a common optical omni-directional receiver (but also in other instances, e.g., RF) the sensor or software structure is configured to be capable of discerning one emitter from another when several signals are simultaneously received. If different frequencies are available, then these may be employed. One strategy is for the emitters to take turns, being synchronized by their communications with the robot or each other. However, this is complex and introduces a failure mode when synchronization breaks down. It is advantageous for each beacon 202, 203 and dock 240 to self-control, and stagger their own signals in time, with intervals between each emission. For instance, each transmitted emission may be differentiated by time division multiplexing (i.e., multiplexing as the beam origins are taken together as a system, including within a beacon having more than one beam or among several beacons) with a fixed period random offset (e.g., which may be different for each beam source). The interval or random offset may be changed from time to time (e.g., at random intervals) or by communication with the robot (e.g., by RF communication when the robot detects an interference condition). At manufacture, or upon changing offset, e.g., the offset may be selected from a set of offsets unlikely to harmonically interfere, or that do not share common factors. In this manner, the robot may encounter several emission paths at once yet discern the identity of each. In other instances, each transmitted emission may be differentiated by different wavelengths of light, infrared modulations, and wavelength filters/windows on the emitters and receivers.

The robot 212 may use RF communication to signal to beacon 202 to remotely activate directed emission 216 and deactivate gateway emission 208 upon initiating the migration mode. In another example, the robot 212 may remotely deactivate the directed vectoring emission 216 and activate gateway emission 208 upon exiting the migration mode.

In some instances, the robot 212 may activate a gateway emission 226 separating bounded areas 206 and 207 and initiate the cleaning mode. Similarly to that described above, the robot 212 may be prevented from leaving bounded area 206 when it encounters gateway emission 208 and 226. When finished cleaning in bounded area 206, the robot 212 may initiate the migration mode and navigate through gateway 228 by following directed emission 230 to a perimeter 234 of a proximity emission 232 and enter bounded area 207. Once in bounded area 207, the robot 212 may re-enter the cleaning or working mode. After, for instance, a set time period or preset number of encounters with gateway emission 208, the robot 212 may migrate from bounded area 204 to bounded area 206.

Alternatively, a virtual wall emitter (gateway) may be set to independently deactivate itself on a schedule to permit the robot to cross the gateway, e.g., acting as a virtual wall for a first interval (to confine in a first or current room), then deactivating temporarily or intermittently, then acting as a virtual wall for a second interval (to confine in a second or successive room). This technique may also be interactive, e.g., the robot may use RF communication with one or more beacons to deactivate a virtual wall or bypass/cross a virtual wall.

It should be noted that in all instances herein, the robot's behavior system is arranged such that obstacle events, including cliff detection events, have a higher priority than any room-to-room navigation or cleaning. Accordingly, for example, when the robot encounters a cliff detection in the middle of a navigation or other set of sequenced behaviors, the robot may nonetheless avoid the cliff (aborting the current part of the sequence and resetting the state of the sequence).

FIG. 23 shows a set of rooms similar to FIG. 22, but arranged (with the addition of a base station) to illustrate the situation in which the robot encounters multiple beams from two beacon emitters and also from a dock 240. Using the strategies avoiding beam confusion noted above, the robot 212 may nonetheless navigate from room to room or the robot 212 may autonomously initiate a docking mode to maneuver towards the base station 240 in bounded area 206 and dock with the station upon termination of the migration mode in the second area.

The base station 240 may include a base, a robot charger, an omni-directional beam emitter and two navigational field emitters and be similar to the base station 180 described above. The robot 212 may maneuver towards base station 240 by detecting and advancing along one of the lateral field edges 242 or 244 of the overlapping fields 246, 248 aligned with the docking direction until docked with the station 240.

The robot 212 may detect the emission of base station 240 with omni-directional receiver 222 on the robot and maneuver to detect an outer lateral field edge (e.g. 247) of at least one field emission 246 or 248. The robot 212 may then advance along outer lateral field edge 247 or 249 to the aligned lateral field edge 242 or 244 of the overlapping fields. Upon detecting the aligned lateral field edge 242 or 244, the robot 212 advances along the aligned lateral field edge 242 or 244 until docked with base station 240.

FIG. 24A is a block diagram showing a behavioral software architecture within a controller 250, in which goal-oriented or sequenced behaviors are depicted in simplified form. FIG. 24B shows the same software architecture, but provides more detail on the goal-oriented or sequenced behaviors while simplifying some of the elements of FIG. 24A.

While embodiments of the robot discussed herein may use behavioral based control only in part or not at all, behavior based control is effective at controlling the robot to be robust (i.e. not getting stuck or failing) as well as safe. Accordingly, FIGS. 24A and 24B depict a software architecture that permits improved control of a mobile robot. Rather than being a purely reactive architecture, FIGS. 24A and 24B introduce an element of planning that is partially reactive (and therefore more robots)—goal oriented, sequenced reactive behaviors.

In the reactive yet sequenced behaviors, the implied success of a related behavior in a sequence is an enabling condition (among others) for each behavior beyond the initiating behavior in the sequence (some sequences may have more than one initiating behavior or branches). The implied success is dept as a state signifying an enabling condition of a successive behavior, in the form of a flag, state machine, or the like. The state may be set or changed by a behavior itself upon completion, by an arbiter process, or by a planning process.

Referring to FIG. 24A, the robot 302 employs a control and software architecture that has a number of behaviors that are executed by an arbiter 252 in controller 250. A behavior is entered into the arbiter 252 in response to a sensor event, or a sequence state event from the planner 253. In one embodiment, all behaviors have a fixed relative priority with respect to one another. The arbiter 252 (in this case) recognizes enabling conditions, which behaviors have a full set of enabling conditions, and selects the behavior having the highest priority among those that have fulfilled enabling conditions. The diagram shown in FIGS. 24A and 24B does not necessarily reflect the (fixed) priority hierarchy of the robot 302. In order of decreasing priority, the behaviors are generally categorized as escape and/or avoidance behaviors (such as avoiding a cliff or escaping a corner), working behaviors (e.g., wall following, bouncing, or driving in a straight line), and for the purposes of this application, sequenced reactive behaviors. The behaviors may include: different escape behaviors 255, 255a, 255n (including escaping corners, anti-canyoning, stuck situations, “ballistic” temporary fire-and-forget movement that suppress some avoid behaviors, e.g., as disclosed in U.S. Pat. No. 6,809,490, the entire contents of the which are hereby incorporated by reference) cliff avoiding 256, virtual wall avoiding 258 (a virtual wall may be a beacon with a gateway beam), other avoidance 258n as a class (e.g., avoiding people via a pyrometer, in general the class of detecting and moving away from something that is a hazard to the robot 302 or to which the robot 302 may be a hazard), spot coverage 264a (covering in a confined pattern such as a spiral or boustrophedon patch), align 264b (turning in place, using side proximity sensors to align with a forward obstacle encountered while obstacle following, e.g., an inside corner), following 264c (representing either or both of substantially parallel or bump following along an obstacle using a side proximity sensor or bumper that extends to the side of the robot), responding to a bump in order to “bounce” 264d (a behavior that occurs after the robot bumps an object), and drive 266. The following is excerpted from U.S. Pat. No. 6,809,490: “Escape behaviors are special behaviors that are given priority when one or more sensor inputs suggest that the robot may not be operating freely.” Movement of the robot 302, if any, occurs while a behavior is arbitrated. If more than one behavior is in the arbiter 252, the behavior with a higher priority is executed, as long as any corresponding required conditions are met. For example, the cliff avoiding behavior 256 will not be executed unless a cliff has been detected by a cliff detection sensor, but execution of the cliff avoiding behavior 256 always takes precedence over the execution of other behaviors that also have satisfied enabling conditions.

The reactive behaviors have, as their enabling conditions or triggers, various sensors and detections of phenomena, but, in general, not (arbitrary) states of a sequence. As shown in FIG. 24A, these include sensors for obstacle avoidance and detection, such as forward proximity detection (multiple), forward bump detection (multiple), cliff sensors (multiple), detection of a virtual wall signal (which may instead be considered a coverage trigger). Sensors of these types are be monitored and conditioned by filters, conditioning 292c, and their drivers 292b, which can generate the enabling conditions as well as record data that helps the behavior act predictably and on all available information (e.g., conversion to one-bit “true/false” signals, recording of likely angle of impact or incidence based on strength or time differences from a group of sensors, or historical, averaging, frequency, or variance information).

Actual physical sensors may be represented in the architecture by “virtual” sensors synthesized from the conditioning and drivers. Additional “virtual” sensors that are synthesized from detectable or interpreted physical properties, proprioceptive or interpreted upon the robot 302, such as over-current of a motor, stasis or stuck condition of the robot 302 (by monitoring a lack of odometry reading from a wheel encoder or counter), battery charge state via coulometry, and other virtual sensors “virtual N.”

In addition, reactive behaviors can act according to enabling conditions that represent detected phenomena to be sought or followed. A beam or wireless (RF, acoustic) signal can be detected without direction; or in some cases with direction. A remote beam or marker (bar code, retro-reflective, distinctive, fiducial, or natural recognized by vision landmark) giving a direction can permit homing or relative movement; without direction the robot 302 can nonetheless move to servo on the presence, absence, and/or relative strength of a detected signal. The reflection of a beam from the robot 302, edge, or line can be similarly detected, and following behaviors (such as obstacle following by the robot 302) conducted by servoing on such signal. A debris or artifact signal can be collected by monitoring debris or objects collected by or traversed by the robot, and that signal can be an enabling condition for a reactive behavior controlling a spot coverage pattern. A general class of “seek/servo N” signals are thus also reactive behavior enabling conditions or triggers.

The robot 302 maintains concurrent processes, “parallel” processes that are not generally considered reactive behaviors. A scheduler 292d may be necessary to allocate processor time to most other processes, e.g., including the arbiter and behaviors, in a co-operative or other multitasking manner. If more threading is available, less processes may be managed by the scheduler 292d. As noted, filters and conditioning 292c and drivers 292b, can interpret and translate raw signals. These processes are not considered reactive behaviors, and exercise no direct control over the motor drives or other actuators. In addition, in the present embodiment, brush motor controller(s) 292a control the main and side brushes, although these may alternatively be controlled by dedicated brush behaviors and a brush control arbiter.

With reference to FIG. 24B (the same representation as FIG. 24A, with other areas of the drawing emphasized/deemphasized for clarity), a special concurrent process is a sequence planner 253. The sequence planner 253 is capable of monitoring and/or maintaining state in order to control sequenced reactive behaviors, e.g., setting enabling conditions (that will be caught by the arbiter 252), and/or resetting states in the case of abort conditions (e.g., time-outs, escape or avoidance behavior execution), and/or monitoring or supervising finite state machines that themselves transition between states. It should be noted that enabling condition states or flags or supervised state machines that request a transition between sequential reactive behaviors can be set by a behavior itself, by the arbiter, or by the sequence planner 253, depending on efficiencies and circumstances. It should also be noted that the sequence planner 253 may optionally be capable of directly controlling the arbiter 252 (or second arbiter 252A) to prefer (e.g., re-prioritize) a behavior or directly control a sequenced reactive behavior without setting an enabling condition, state variable or state machine.

While maintaining state is recognized to potentially reduce the robustness of competent reactive control architecture, sequences of intended actions may require the maintenance of some state. As discussed herein, if sequenced reactive behaviors are controlled to execute using request states which may be robustly reset or erased without primary function failure, the reduction in robustness can be reduced.

One exemplary sequenced reactive behavior set is employed by the robot 302 to transition between rooms or navigate to a distal room. As depicted in FIG. 24B, a homing beam align behavior 294a includes an enabling condition that an align start (e.g., a homing request) enabling condition be activated, and that the beam to be homed upon is detected by a multi-directional sensor. As will be discussed below, the homing request enabling condition can be set and maintained indifferent ways. A homing beam homing behavior 294b is to execute once alignment is successful. However, because alignment may happen serendipitously (the robot may already be aligned when the homing request is activated), the homing beam homing condition also has as an enabling condition the homing request (flag or state), as well as a detection using a directional sensor (binoculars) that substantially necessarily directed at the beam identified as one to be homed upon (or, e.g., filtered to be equivalent to such a detection). Accordingly, two sequenced reactive behaviors are each separate entry points to the sequence, and share a common reactive sequence request enabling condition. When a sensor-based enabling condition fails, these behaviors are aborted (internally or by the planner 253) and the sequence request remains active (or may be reset). In this case, the sequence may be restarted the next time either sensor-based enabling condition occurs. Optionally, a random component (such as a random distance or time return to random coverage, while disabling the sequence request or homing request enabling condition) is permitted to occur following an abort of a sequential reactive behavior in order to decrease the likelihood of systematic failure or looping.

After the homing beam homing behavior 294b has completed, the sequence is in a state where an assumption is made regarding the state that the robot is located appropriately to align with the field to permit begin field following about a beacon to transition to a new chamber. A (new) align start enabling condition (e.g., field align state or request) may be set by the completing sequenced reactive behavior or by the planner, via condition flag or state machine. In order to accommodate the possibility that the field may be the first detectable signal encountered upon initiation of the homing request, the detection of the proximity field by the multi-directional receiver is an equivalent (OR) condition to the field align request, if the homing request is active when the field is encountered, this combination also should initiate field align behavior 294c. If there are no abort conditions, the field align behavior 294c executes. After the field align behavior 294c has completed, the sequence (of sequenced reactive behaviors) is in a state where a next assumption is made regarding the state—that the robot is oriented appropriately to begin field following. A (new) follow start enabling condition (e.g., field follow state or request) may be set by the completing sequenced reactive behavior or by the planner, via condition flag or state machine. Similarly, the launch behavior 294e may have as a primary enabling condition a state-based request rather than a sensor detection. Each of these behaviors 294a, 294b, 294c, 294d, 294e is detailed herein below.

In two of these cases, optionally, a state to enter the entire sequence (homing request) is maintained and an additional state to begin a next part of the sequence (field follow request; launch request) is set and both conditions are necessary to start the next sequenced reactive behavior. This would permit restarting the sequence immediately after an abort (by maintaining the entire sequence request state). Alternatively, the entire sequence entry request is reset and the only the next sequenced reactive behavior entry condition is set (field align request or field follow request). This would cause the entire sequence to be otherwise restarted (e.g., by a new or continued occurrence of conditions that would cause the planner to set the sequence request) after an abort, which could improve general robustness (although not necessarily goal-specific robustness) by erasing essentially all states.

In two cases, no sensor-based condition is an enabling condition for the next sequenced reactive behavior in the sequence, but an enabling condition is a sequence-based or a state-based request—e.g., field follow request, launch request. Accordingly, a sequence may include sequenced reactive behaviors that include only sequence states or progress as enabling conditions for the next sequenced reactive behavior (such as the field follow behavior, or launch behavior), as well as sequenced reactive behaviors that include both sensor-based conditions and sequence states, progress or requests as enabling conditions (such as the beam align or tractor homing behaviors).

In one case, a state to enter the entire sequence (homing request) is maintained and either a sensor-based condition or a sequence-based request is an enabling condition for a next sequenced reactive behavior (field align) in the sequence—e.g., a homing request AND (field align request OR field detection) is interpreted by the arbiter to permit field aligning if no reactive behaviors have higher priority. Accordingly, a sequence may include sequenced reactive behaviors that include either a combination of sequence state/requests, or both sensor-based conditions and sequence request/states, as enabling conditions for a next sequenced reactive behavior.

It should be noted that entry and mid-sequence branching is possible by enabling different behaviors depending on different combinations of sensor-based conditions and sequence states. For example, just as two possible entry points into the sequence are permitted by detecting different sensor events (multi-directional or directional) at the same time as a common entry point enabling condition (homing request), branches in the sequence to different next sequenced reactive behaviors are permitted by detecting different sensor events at the same time as a common mid-sequence enabling condition (e.g., candidate behaviors each enabled by a unique combination of sensor event and sequence state). Optionally, if different sequences each are maintained with different states, different mid-sequence state combinations permit nesting of sequences (candidate behaviors each enable by a unique combination).

Sequence planner 253 optionally maintains a navigation mode state switch or finite state machine 243a, and/or such states or requests are maintained separately (e.g., follow start, follow done).

As shown in FIG. 24B, the planner 253 may control different goal-oriented sequences 1-N, each having more than one sequenced reactive behavior therein. The behaviors therein can have as initial conditions a goal state, entry request, or other state, that the arbiter 252 identifies as requiring rendering a candidate behavior available for execution. Initial behaviors of a sequence may require a sequence request; subsequent behaviors may require either a next sequenced reactive behavior request (or, in addition, the sequence request as well).

The sequenced behaviors are “reactive” despite being dependent upon sequence state at least because they may be subject to the same arbiter 252 as the reactive behaviors, and/or react to sequence states as enabling conditions in the same manner as reactive behaviors react to real-time sensor events, and/or are prioritized lower than at least escape and avoidance behaviors. In addition, the states may be wholly reset to restart a reactive process. Even when a sequence or next sequenced reactive behavior's enabling conditions are satisfied, a reactive behavior that also has enabling conditions satisfied and preserves continuous autonomous operation will be executed instead. Reactive docking behaviors, which generally permit the robot to return and recharge for continued autonomous operation, may be prioritized higher than, in particular, navigation-related sequenced reactive behaviors. Reactive coverage behaviors may be prioritized below navigation sequenced reactive behaviors, in particular. Some portions of the docking behaviors noted herein are alternatively arranged as sequenced reactive behaviors; or other sequences are possible. Beacon-based navigation is discussed herein is an example of one implementation of combining sequenced reactive behaviors with non-sequenced reactive behaviors.

As shown in FIG. 24B, a second arbiter 252a may supervise a sequence of sequenced reactive behaviors (e.g., in combination with the sequence planner 253); or a master arbiter (not shown) may supervise a first arbiter (e.g., 252) of reactive behaviors and a subsequent arbiters (e.g., 252a) of different sequenced reactive behaviors.

As shown in FIG. 24B, for a navigation application in particular, the sequence planner 253 may include, monitor, or maintain location state in the form of a topological state map or its equivalent with respect to the environment in which the robot operates. Using beacons, this may be a topological map having nodes and links, in which the nodes may be rooms (links thereby being gateways or beacons) or gateways (links thereby being sides of the gateway). A primary use of the location state is to provide a homing beam or homing fiducial identity to the arbiter or to the sequenced reactive behaviors. If provided to the arbiter, the arbiter may recognize a sensor-based enabling condition (e.g., beam detection) for a behavior only when the beam has the correct identity (encoding) for the location state. Alternatively, the behavior itself examines the identity or encoding of a received enabling condition (e.g., beam detection). An abort condition can result if beacons have been rearranged or the like, and this abort condition can cause blacklisting of a bacon or fiducial or a reset of the map (new mapping needed). In either case, the robot 302 continues to provide coverage and all other remaining reactive behaviors in the local area. The location state is used by the planner 253 to successively execute sequences of sequenced reactive behaviors in order to traverse a distance to or from a remote or distal room (for returning from the dock for charging; for directly proceeding to a remote or distal chamber from the dock); and permits the same sequenced reactive behaviors to be used for navigation in different directions (so long as the location state is maintained) by providing the identities of the guiding fiducials or beams in the correct order for a navigation direction, depending on the direction to be traveled on the topological map.

One candidate routine maintained/executed by the planner 253 is shown in FIG. 25A. The elements of this routine are not necessarily in order (e.g., different states may be checked out of the depicted order). The planner routine monitors (502) a goal-oriented or sequence initiating condition. As discussed, this may be a variable that keeps track of a time, distance, cleanliness measure, history of obstacle interactions, or other criterion that, when satisfied, indicates that a planned sequence (such as navigation from room to room) is to be carried out. The condition may be a state, flag or stare of a finite state machine. If such a condition exists (504), the planner checks (506) the state of the robot along the called for sequence, if the planner maintains such a sequence state. For example, the planner 253 may maintain or monitor the current state of a state map or finite state machine defining states of the beacon crossing sequence, as shown in FIG. 25B (in which “aligning to launch”, “homing”, “aligning after homing”, and “following field” states are available); alternatively a state map or finite state machine may be instantiated. The planner 253 may additionally or in the alternative set (508) behavior enabling conditions for entry into a sequence, such as a homing request ad discussed herein, or an entire sequence entry behavior enabling condition and/or next sequenced reactive behavior enabling condition. The planner 253 may additionally set (510) guiding variable for the behavior or arbiter to recognize (e.g., such as the identity of a beam to be followed or fiducial to be homed upon or bearing to proceed in), according to, e.g., the state of a map identifying such (e.g., as shown in FIG. 25C, according to the state of a map defining which beam A or B—“green” or “red”—of lighthouses/beacons X, Y, Z connects to which rooms 1, 2, 3 and a dock, and according to the next goal room defined by the planner 253). The planner 253 may additionally monitor abort conditions (512, e.g., time out, interruption by a reactive avoidance or escape behavior) of sequenced reactive behaviors in progress, and set or reset the state of a sequence and/or the state of a map upon such an abort (although behaviors may also set or reset these internally before aborting).

After the robot 302 has detected a specified cleaning period has elapsed or the robot 302 has detected a specified number of bumps or the robot has detected a specified condition requiring a set of sequenced behaviors, the sequence planner 253 executing on the controller 250 will decide to exit the cleaning mode and enter the migration mode. The robot may also enter the migration mode right from the dock, in order to proceed to a distal or remote room in order to begin cleaning there. The migration mode executes the task of moving from one room to an adjacent room. As discussed above, the migration mode is advantageously not a monolithic process, but is a succession of sequenced reactive behaviors started by an initial behavior—that has an enabling condition of a migration mode request (also discussed herein as a “homing request”). The planner 253 will raise a “new room”/homing request enabling condition, state, or flag. The remaining enabling condition is a sensor detection of a guiding tractor beam (optionally, limited to one beam identified according to the location state 253b as originating with the lighthouse/beacon to be homed upon), which, when the beam is detected (substantially immediately if the robot is in the beam, or after the robot encounters the beam as it continues the coverage behaviors having satisfied enabling conditions), triggers the homing beam—align behavior 294a to be selected by the arbiter 252 for execution. The robot 302 performs the task of migrating from one bounded area to an adjacent bounded area by executing the homing beam—align behavior 294a while at the same time possibly executing the cliff avoiding 256 (on any occasion when any cliff detector fails to see the floor—i.e., the enabling condition of the higher priority & reactive avoidance behavior is also satisfied) and drive 266 behaviors (e.g., if an abort condition resets the sequence). When multiple behaviors are trigged by sensor input or the planner 253, the arbiter 252 will arbitrate the behaviors in order of priority. That is, the arbiter may also arbitrate among different sequences of reactive behaviors, or the planner may only start one sequence. The beacon crossing sequence 262 has five sequenced reactive behaviors, beginning with the homing beam—align behavior 294a, which are executed or sequentially based on controller inputs as discussed herein. If the robot 302 experiences a signal input (such as a bump sensor input) that requires execution of a behavior having a higher priority, the robot 302 will execute the behavior of higher priority and afterwards, upon the presence of the necessary enabling conditions, may reinitiate a previous sequence of reactive sequenced behaviors.

FIGS. 31A-G are overhead views showing a behavior based autonomous coverage robot 302 as it uses a navigation beacon 304 to migrate from one area to another. FIGS. 26-30 are diagrams illustrating the beacon crossing sequenced reactive behaviors 262, including beam aligning 294a, beam homing 294b, field following 294c, field aligning 294d, and leaving beacon 294e behaviors.

The beacon crossing sequence 262 commences by entering the beam homing 294b or beam aligning behaviors 294a, depending on a robot position within the first bounded area in relation to the beacon 304. Generally, the robot 302 will initiate a maneuver to detect with its omni-directional receiver 104 the first directed beam 306 from the navigational beacon 304 into a first area (e.g. turning 360 degrees or driving in a circle).

FIG. 26 depicts the beam align behavior 294a. As shown in the arbiter 252 will not permit the beam align behavior 294a as a candidate (i.e., to be prioritized with other behaviors with satisfied enabling conditions) at least until a homing request state or request is present, as well as the directional or tractor beam 306 is received by the multi or omni-directional sensor 322 of the robot 302, as behavior enabling conditions or triggers. To permit execution, no higher priority behaviors (e.g., reactive behaviors) have satisfied enabling conditions. This tractor or directed beam 306 may be limited to one expected at the present location of the robot 302 on the state map (e.g., by responding only to a beam having an identity or encoding corresponding to the current map state). Once these are satisfied, the arbiter 252 permits the beam align behavior 294a to execute. If either condition is no longer satisfied, the behavior may abort. As noted in FIG. 26, abort conditions (e.g., including missing behavior enabling conditions; time-out; or displacement by a higher priority behavior) may be monitored by the planner 253, which may also keep states reflecting the progress of the sequence as well as the location of the robot 302. The beam align behavior 294a turns or pivots the robot 302 essentially in place (522), starting in a random direction. A likely beam direction may also be maintained as a state, e.g., based on past beam detections monitored by, e.g., a concurrent process, and if so, the turning direction may be the one expected to be shorter. Once the directional receiver 318 senses the directed beam 306 (524), the behavior ends. There is no need to set a state as a request for the beam homing behavior 294b likely to follow, as the robot 302 will, so long as the homing request is active, continue to attempt to align and home when the correct beam is encountered and either is a valid entry point for the sequence. Later stages of the sequence are mid-sequence, and may use such states. Accordingly, the beam align behavior 294a is an example of choosing not to set a state in order to maintain a higher level of reactive control and likely robustness.

FIG. 31A shows the robot 302 during readiness for and execution of the beam align behavior 294a, as it encounters a first directed beam 306 emitted by a navigation beacon 304, when a homing request from the planner 293 is active. Upon detecting the first directed beam 306, the robot 302 commences the beam aligning behavior 294a. During beam aligning 294a, the drive system 130 maneuvers (e.g., by turning in place, e.g., starting in a random direction) the robot 302 to detect the first directed beam emission 306 with the directional receiver 318, which is to be aligned with a robot drive direction. By using the signals from the two component receivers 114, 116 of the directional receiver 318, the robot 302 aligns the drive direction with the directed emission 306 of the beacon 304. FIG. 31B shows the robot 302 as it detects beam 306 with the omni-directional receiver 322 and turns to face beacon 304.

After the robot 302 has aligned the drive direction with the directed emission 306 of the beacon 304 by using the directional receiver 106, the robot may commence the beam homing behavior 294b.

FIG. 27 depicts the tractor or directed beam homing behavior 294b. As shown, the arbiter 252 will not permit execution (of the highest priority candidate) until at least a homing request state or request is active, and the directional or tractor beam 306 is received by the directional sensor 318 of the robot 322 as behavior enabling conditions or triggers. This tractor or directed beam 306 may again be limited to one expected at the present location of the robot 302 on the state map. Once these are satisfied, the arbiter 252 may permit the beam homing behavior 294b to execute. If either condition is no longer satisfied, the behavior 294b may abort. As noted in FIG. 217, abort conditions and resulting states may be supervised by the planner 253. The tractor or directed beam homing behavior 294b follows the directed beam emission (532). One exemplary way of following is using alternating curves. Each parallel sensor 114 and 116 of the directional binocular sensor 318 is conditioned to return a one bit true-false value reflecting beam presence, based on signal strength (e.g., independently thresholded, or thresholded with respect to one another). As the robot 302 turns toward and away from the beam 306, beam detections of true-false, true-true, and false-true are possible (conditioning may also limit the possibilities to true-false and false-true). A “bang-bang” servoing may be used to change turning direction as the robot 302 moves forward. Each turn is an arc with decreasing radius, i.e., part of an inward spiral. Alternatively, analog values may be maintained and more complex feedback control employed. The following (532) may alternatively employ the frequency and variance tracking discussed herein with respect to docking to determine that the robot is proceeding in the correct direction and move forward in a straight line. Once the non-directional/omni-directional receiver 322 senses the directed beam 306 (534), the behavior is ready to end. A state is to be set to continue the sequence of sequenced reactive behaviors—i.e., a field aligning request. The behavior 294b may set this state itself (536), e.g., as a flag in memory; or the planner 253 may monitor that the behavior 294a is completed and set such a flag or state in a finite state map or machine.

The behavioral sequence, as discussed, permits the robot 302, after initiating the beacon crossing sequence 262, to skip the beam aligning behavior 294a if the directional receiver 318 already detects the directed beam emission 306 and is already aligned with the directed beam emission 306. Abort conditions may present these and permit later restarts. As discussed herein a combination of reactive behavior based control and sequenced reactive behavior control is advantageous. However, some embodiments discussed herein do not necessarily exclude other reactive or non reactive techniques. While executing the beam homing behavior 294b, the robot 302 moves towards the navigational beacon 304 while maintaining its alignment with (e.g., servoing upon) the directed beam 306 in response to the signals of the directional receiver 318. FIG. 31C shows the robot 302 approaching the navigation beacon 304. As the robot 302 follows the edge 310 of the beam 306, the robot 302 adjusts it course as shown by arrows 308. As shown in FIG. 31C, the robot 302 stops moving towards the beacon 304 when it encounters a proximity field emission (or “force field”) 310 projected laterally from beacon 304 in all directions.

As an alternative behavior that would be suitable as part of the sequenced reactive behaviors, if the directional receiver 318 is selected to be more sensitive (e.g., by collimation and confinement) at remote distances than the omni-directional receiver 322, the directional receiver 318 may receive the force field emission 310 even though the omni-directional sensor 322 does not. In this case, the directional sensor 318 may be used to alternatively home on the force field emission 310. In this case, an alternative behavior is provided in the sequence, identical to the tractor homing behavior 294b except that the sensor-based enabling condition is that the force field 310 is seen in the directional sensor 318; and the homing is on this emission 310. As an example of behavioral robustness, this behavior may be arbitrarily added to the sequence of reactive behaviors without changing any other behavior or disrupting any other process, with a priority above or below the existing tractor homing behavior 294a depending on empirical evidence that one or the other tended to be more reliable.

Upon detecting the proximity field 310, the robot 302 executes the field aligning behavior 294c. FIG. 28 depicts the field align behavior 294c. As shown, the arbiter 252 will not permit execution (of the highest priority candidate) until at least a homing request state or request is active, and either a field align request is active, or the proximity field 310 is seen in the non-directional sensor 322 (this latter permitting the behavior sequence to advance to field aligning 294c when the robot 302 fortuitously “stumbles” into the proximity field 310 before encountering the directed beam 306). The field may again be limited to one expected at the present location of the robot 302 on the state map. Once these are satisfied, the arbiter 252 may permit the field align behavior 294c to execute, abort conditions supervised as previously noted by the planner 253. The field align behavior 294c turns or pivots the robot essentially in place (544). A likely beam direction may be determined according to the beam identity (e.g., facing the dock 340, following a left beam 350 to the dock 340 likely means the robot 302 should turn clockwise, following a right beam 360 counterclockwise) or other state or information. The robot 302 may turn in the direction expected to be shorter or randomly. Once the omni or non-directional receiver 322 senses the directed beam 306 (546), the behavior 294c ends. Optionally, the behavior 294c does not end until an additional condition is satisfied—that the directional beam 306 is not present in the directional receiver 318 (which increases the likelihood that the robot 302 is rotated transversely to the dock 340). A field follow request is to be set to continue the sequence of sequenced reactive behaviors, as discussed, by the behavior itself (548), by the planner 253, or otherwise.

Accordingly, as shown in FIG. 31D, the robot 302 can detect the proximity field 310 with the directional receiver 318, as when the robot 302 is facing the beacon 304, the robot 302 will initiate a pivot maneuver (e.g., turn in place for differential drive robots having drive wheels on the center diameter or holonomic drives or highly steerably robots, turn substantially in place for other configurations) altering the robot drive direction to the right or left toward the other directed beam 314, based on the beam encoding of the first directed beam 306. The robot 302 stops the pivot maneuver when it no longer detects the proximity beam 310 with the directional receiver 106.

The robot 302 may encounter the proximity field 310 before encountering the directed beam 306. In this case, the robot 302 does not execute the beam aligning or beam homing behavior, 294a and 294b, respectively. Instead, the robot 302 recognizes the encoding of the proximity field 310 and executes the field aligning behavior 294d.

When the robot 302 no longer detects the proximity field beam 310 with the directional receiver 218. The robot 302 executes the field following behavior 294c.

FIG. 29 depicts the field following behavior 294d. As shown, the arbiter 252 will not permit execution (of the highest priority candidate) until at least a homing request state or request is active, and a field follow request is active. As noted, this may be limited to simply the field follow request if the sequence is to be more readily abandoned to preserve robustness. The field is optionally limited to an expected one. Once these are satisfied, the arbiter 252 may permit the field follow behavior 294d to execute subject to abort conditions. The field follow behavior 294c follows the edge 312 of the proximity field emission 310 (554) using alternating, decreasing radius curves as previously discusses. The non-directional sensor 322 may be conditioned to return a one bit true-false value reflecting beam presence, based on signal strength. Once the non-directional/omni-directional receiver 322 senses the next directed beam 314 (534), i.e., the launch beam (this detection limited to the expected beam of the beacon 304 followed and/or according to the location state of the robot 304), the behavior is ready to end. Optionally, the behavior 294d further checks and proceeds until the directional receiver 318 does not sense the launch beam 314 (which tends to exclude those cases in which the omni-directional detector 322 detects a launch beam 314 too early via a reflection from a highly reflective surface, such as a nearby white door or wall corner). A state is to be set to continue the sequence of sequenced reactive behaviors—i.e., a launch request. The behavior 294c may set this state itself (558), or the planner 253 may control this.

As shown in FIG. 31E, during the field following behavior 294d, the robot 302 maneuvers in an resultant are around the beacon 304 while following an edge 312 of the proximity beam 310 via omni-directional receiver 322. The robot 302 follows the edge 312 of the force field 310 by monitoring the received signal strength of the force field emission 310 on the robot's emission sensors 318 and 322. One example of the arc maneuver entails the robot 302 “zigzagging” back and forth (e.g., “band-bang” servoing or following using a one-bit true-false detection and forward motion in initially arced, steadily decreasing turn radius, e.g., spiraling in) along the edge 312 of the proximity beam 310, where the robot 302 detects entering and exiting the field of emission of the proximity beam 310 with the omni-directional receiver 322. In another example, the robot 302 follows the edge 312 of the force field by using a PID control system based on the time in the force field 310 or the time out of the force field 310. The PID control system controls the arc of the robot 302. The robot 302 executes the maneuver at a reduced speed (e.g., less than 50% of coverage speed) for both drive wheels 132 and 134, so that both drive wheels 132 and 134 are being driven in a forward direction. This facilitates easier traversing of transitions and thresholds. The robot 302 passes through the area normally covered by a gateway beam 316. The robot 302 continues to follow the edge 312 of the proximity beam 310 until the omni-directional receiver 322 detects that the robot 302 has entered the second directional beam 314 and the directional receiver 318 no longer detects the second directed beam 14, as shown in FIG. 31E. As noted, satisfying both criteria helps prevent the robot 302 from detecting reflected beam emissions off of adjacent objects, such as a white door.

FIG. 29 depicts the launch behavior 294e. As shown, the arbiter 252 will not permit execution (of the highest priority candidate) until the launch request is active, and may also require the homing request. The launch behavior 294e assumes a last position with the omni-directional detector 322 in the launch beam 314 and a bearing substantially transverse to the launch beam 314, and curves away from the field 314/beacon 304 in the approximate direction of the launch beam 314 (e.g., facing the dock 340, if the robot 302 has traversed left to right, curving clockwise, and if the robot 202 has traversed the opposite direction, curving counterclockwise). Optionally, the robot 302 may try to stay within the launch beam 314 by servoing to return to the launch beam 314 when the launch beam 314 is not detected in the omni-directional sensor 322. After a predetermined time (e.g., 5 s) or distance (e.g., 1 m) or obstacle encounter, the behavior is ready to end. At this point, the planner 253 may terminate the homing request (and, e.g., updating the location state of the topological map as well as clearing or resetting a navigation mode state is one is kept), and the reactive behaviors will begin the robot's coverage work in the subsequent chamber (or, if the battery is to be charged, this may initiate another homing request in the direction of the dock according to the location state of the map). The planner 253 may also immediately initiate another homing request; or reset variables of the current homing request according to the location state of the map such that the robot 302 will proceed from room to room to room, e.g., moving directly to or from a distal or remote room.

That is, as shown in FIG. 31F, upon completing the field following behavior 294c, the robot 302 may execute the leaving beacon behavior 294e, as shown in FIG. 30, which entails the robot 302 away from the beacon 304. Alternatively, this can be performed as a pivot and launch, and/or guided by the launch beam 314 and omni-directional receiver 322, e.g., FIG. 31G shows the robot 302 turning away from navigation beacon 304 to face a second area as the robot 302 passes the edge of directed beam 314. If the robot 302 detects the second directed beam emission 314 with the omni-directional receiver 322, the robot 302 maneuvers away from the beacon 304. If the robot 302 does not detect the second directed beam omission 314 with the omni-directional receiver 322, the robot 302 moves in a forward direction as shown in FIG. 31H without turning in order to avoid arcing back into the gateway. Another example of the leave beacon behavior 294e includes the robot 302 aligning itself in a particular direction using odometry (or a next beacon in a sequence e.g., when the planner 253 includes a distal or remote room traversal sequence that restarts the sequence of behaviors at beam align with a next beacon's beam or field according to the map) and moving in that direction without guidance from any of the beams emitted from to current beacon 304. FIG. 31G shows the robot 302 moving away from navigation beacon 304 as it finishes migrating.

While migrating, the robot 302 stores in memory the beam encoding of the first and send directed beams 306 and 314, respectively, at each navigational beacon 304 encountered while away from a base station 340. To return to the base station 340, the robot 302 conducts coverage behaviors while the homing request is on, resulting in discovering the appropriate directed beams 306 and 314, respectively, and corresponding encodings, to follow a sequence of encountered beams back to the base station 340. In one example the robot 302 can wirelessly communicate (e.g. radiofrequency) with the navigational beacons 304 to independently activate each previously encountered beam 306, 314, and 316 one at a time to find (traversing a room using random bounce and/or obstacle following to place itself in a beam path) and follow each beam back to the base station 340.

In another example (not shown), navigation beacons may emit at least one, two, or three beam signals; fence/gate beam (referred to herein as “yellow”, but in reality having a first modulation of a first frequency in the IR band), right-hand navigation beam (referred to as “red”, but in reality having either or both of different modulation or a different frequency in the IR band from the “yellow” beam), and left-hand navigation beam (referred to as “green”, but in reality having either or both of a different modulation or a different frequency in the IR band from the “yellow” and “red” beams).

If multiple navigation beacons are provided, each would have different modulation of its beams, but the emitters of the fence/gate beam, right hand beam, or left hand beam would preferably be consistent among the navigation beacons.

Regardless of its assignment as fence, gate, or trail marker, as one example, a navigation beacon also may emit a geometrically determined force field (“blue”, again in actuality either or both of a different modulation or frequency in the IR band, although a force field could also be radio frequency RF rather than visible or IR light) in a circular region around their base. This may identify its presence to a robot, and allow a robot to follow a path around the Navigation beacon (through a gate or past a trail marker), for example.

An IR-based beacon system can be designed to use a geometric solution, similar in certain respects to a non-limiting example such as a shaded floor lamp which lights up only a small area around itself, in order to provide a well-defined force field via IR, for example.

A fence navigation beacon may emit a yellow beam directly out the front of the navigation beacon (0 degrees) and a single navigation beam (red or green) at about a 60-degree angle to the right or left of the yellow beam, for example. A gate navigation beacon may emit a yellow beam to the front a red beam at −60 degrees, and a green beam at about +60 degrees, as another non-limiting example; also, for example a trail marker Navigation beacon may emit a red beam at about −60 degrees and a green beam at about +60 degrees.

Navigation beacon “beams” in some non-limiting examples may not be narrowly focused (thus not requiring a special lens); they may fan out to light as much of the room as may be practicable without mutual interference. As an alternative, for example, a standard home base LED not requiring other optics may also be used.

Referring to FIGS. 24A-B and 32A-E, the docking behavior priorities (from highest to lowest) include docking retry docking bump follow 264c, docking bounce 264d, docking quick turn 296c, docking distant homing 296b, docking lobe following 296a, and docking field following 296d. The robot 302 must generally approach the dock 340 from a forward direction with less than 4 degrees of skew in order to dock properly.

The default docking behavior, docking lobe following 296a, as shown in FIG. 32A, is executed when no other docking behavior has a higher priority. When the robot 302 detects the left (red) beam 350 only, it follows the edge of the red beam 350 in a clockwise direction. When the robot 302 detects the right (green) beam 360 or the overlap area of the red and green beams 350 and 360, respectively, it follows the edge of the green beam 360 in a counterclockwise direction. This results in the robot 302 appearing to follow the outside of the nearest docking beam 350 or 360 around to the front of the dock 340 at zero degrees and then following a normal (zero degree) edge 362 right onto the docking contacts. When the robot 302 detects the force field 370 during this behavior it slows down in order to follow more accurately.

The robot 302 keeps track of the signals detected by the omni-directional receiver 322 over a window of 30 seconds. When the robot 302 detects that the frequency and variance of line crossing events (crossing the normal (zero degree) edge 362) is above about 1 Hz and below a variance of about 2.5 seconds it determines that the robot 302 is following the normal (zero degree) edge 362 and will simply drive straight by executing a smooth homing behavior, instead of continuing to follow the edge 362 (with the inevitable right-left oscillation) under the lobe following behavior 296a.

Referring to FIG. 32B, the quick turn behavior 296c is activated when the robot 302 has encountered a docking beam 350 or 360 while driving straight, spiraling, bump following, or wall following, is not in the force field 370, and has not detected the dock 340 with its directional receivers 318 in the last several seconds. When the robot 302 detects the red beam 350, it arcs clockwise toward the dock 340. When the robot 302 detects the green beam 360 or the overlap area of the red and green beams 350 and 360, it arcs counterclockwise toward the dock 340. When the robot 302 detects the other beam 350 or 360, detects the dock 340 with its directional receivers 318, or has arced more than 360 degrees, this behavior 296c aborts. Typically it is then followed by the lobe following behavior 296a.

Referring to FIG. 32C, the distant homing behavior 296b is activated when the robot 302 has not detected the force field 370 in its omni-directional receiver 322 in the last several seconds, and detects the dock 340 with the directional receiver 318. If the robot 340 detects the red or green beams 350 and 360, respectively, it will drive toward them. If the robot 340 only detects the force field 370 it will drive toward the force field 370. This allows the robot 302 to approach the dock 340 from a distance at any angle. When the robot 302 detects the force field 370 with its omni-directional receiver 322 this behavior 296b aborts. During a front approach this is typically followed by the lobe following behavior 296a. During a side approach this is typically followed by the field following behavior 296d.

Referring to FIG. 32D, the field following behavior 269b activates when the robot 302 detects the force field 370 with its omni-directional receiver 322, has not detected the red or green beam, 350 and 360 respectively, with its omni-directional receiver 322 in the last second, and does not see the red or green beam 350 and 360, respectively, with its directional receivers 318. The robot 302 follows the edge 372 of the force field 370 with its omni-directional receiver 322 in a random direction (cw or ccw) until it detects the red or green beam, 350 and 360 respectively, a bump, a cliff, or has traveled more than 1 meter. When the robot 302 detects the red or green beam, 350 and 360 respectively, it engages in reverse lobe following 296a for a distance of about 1 meter or until a bump is detected tin order to travel away from the dock 340 along the normal (zero degree) edge 362 and straighten out. Then the robot 302 turns 180 degrees or until it faces the dock 340. Then the behavior 296b aborts. Typically the lobe following behavior 296a activates next to finish docking.

Referring to FIG. 32E, the robot 302 can dock successfully even when there are obstacles completely or partially occluding the docking beams 350 and 360 or the robot's physical access to the dock 340 from some, but not all, approach angles by executing an obstacle dock behavior 296c. When the robot 302 is seeking the dock 340 and had detected a docking beam 350 or 360 in its omni-directional receiver 322 in the last several seconds, and then also detects a bump, it activates the docking bounce behavior 264c. With about a 66% probability, the obstacle dock behavior 296e initiates a bump follow behavior 264c and with about a 33% probability, the docking bounce behavior 264d which has the robot 302 simply back up, turn a random angle between 10 and 170 degrees for side impacts or 75 and 255 degrees for front impacts, and then aborts the behavior 264d. The bump follow behavior 264c follows an edge of the obstacle with the robot's bumper until one of the following conditions occurs: a) the robot 302 detects the dock 340 with the directional receiver 318 and does not detect it with is omni-directional receiver 322, with about a 66% probability; b) the robot 302 detects a crossing of the normal (zero degree) edge 362 from right to left and is bump following 264c in a clockwise direction, or the robot 302 detects a crossing of the normal (zero degree) edge 362 from left to right and is bump following 264c in a counterclockwise direction, with about a 66% probability; c) the robot 362 detects the force field 370 with its omni-directional receiver 322; d) 30 seconds has passed since the onset of the bump follow behavior 264; or e) no docking beam 350 or 360 has been detected by the omni-directional receiver 322 for more than about 5 seconds.

Another possible implementation of bump follow 264c includes the robot 302 keeping track of the projected location of the dock 340 and turning preferentially toward the estimated dock location during bump follow 264c. Each time the robot 302 crosses the normal (zero degree) edge 362 while facing the dock 340 and detecting the dock 340 by its directional receiver 318, the robot 302 uses odometry to project the location of the dock 340 about 5 feet out along a direction of travel. The robot 302 can use odometry throughout the docking maneuver to estimate the orientation of itself to the projected location of the dock 340.

The combination of short bouts of bump following and normal docking methods allow the robot 302 to dock in the face of a large variety of obstacles including but not limited to walls, chairs, and boxes, as well as reflection of the IR docking beams.

“ROBOT OBSTACLE DETECTION SYSTEM”, U.S. Pat. No. 6,594,844, disclosing proximity sensors such as cliff sensors and wall following sensors; “AUTONOMOUS FLOOR-CLEANING ROBOT”, U.S. Pat. No. 6,883,201, disclosing a general structure of an iRobot Roomba coverage/cleaning robot and main and edge cleaning heads in detail; “METHOD AND SYSTEM FOR MULTI-MODE COVERAGE FOR AN AUTONOMOUS ROBOT”, U.S. Pat. No. 6,809,490, disclosing motion control and coverage behaviors, including escape behaviors, selected by an arbiter according to the principles of behavior based robotics; and “METHOD AND SYSTEM FOR ROBOT LOCALIZATION AND CONFINEMENT”, U.S. Pat. No. 6,781,338, disclosing virtual walls, i.e., robot confinement using wall-simulating directed beams, are each incorporated by reference herein in their entireties.

Other robot details and features combinable with those described herein may be found in the following U.S. patent applications, entitled “COVERAGE ROBOT MOBILITY” having assigned Ser. No. 11/633,885; “MODULAR ROBOT” having assigned Ser. No. 11/633,886; and “ROBOT SYSTEM” having assigned Ser. No. 11/633,883, the entire contents of the aforementioned applications are hereby incorporated by reference.

A number of implementations have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the following claims. Accordingly, other implementations are within the scope of the following claims.

Claims

1. A robot comprising: a chassis;a wheeled drive supporting the chassis and operable to maneuver the robot;sensors responsive to an environment about the robot;a memory storing: instructions configured as a set of routines to cause the wheeled drive to move the robot according to a predetermined sequence based on enabling conditions that must be satisfied, the set of routines comprising a plurality of working routines and a plurality of escape routines, wherein each routine of the set is assigned a priority and one or more enabling conditions, andinstructions to enable control of the wheeled drive according to the assigned priorities and enabling conditions, the instructions configured to control the wheeled drive according to a highest priority routine of the set for which all assigned enabling conditions have been satisfied; anda controller in communication with the wheeled drive, the sensors, and the memory, the controller executing the instructions to: identify, from among the set of routines, a subset of routines for which all assigned enabling conditions have been satisfied, the subset of routines comprising one or more working routines and one or more escape routines;select, from among the identified subset of routines, an escape routine having the highest priority of the subset of routines; andcontrol the wheeled drive according to the selected escape routine.

2. The robot of claim 1, wherein routines of the set having a sequence-based enabling condition have a lower priority than each of the one or more escape routines and at least one avoidance routine.

3. The robot of claim 1, wherein each routine of the set effects movement of the wheeled drive.

4. The robot of claim 1, wherein the enabling conditions comprise at least one sensor-based enabling condition satisfied upon a specified sensor detection and at least one sequence-based enabling condition satisfied upon a completion of a specified routine in the predetermined sequence.

5. The robot of claim 4, wherein the set of routines has a sequence-based enabling condition from a parent set of routines, the set of routines being executed in the predetermined sequence upon enablement and a predetermined sequence of the parent set of routines being resumed upon either completion of the set of routines or cessation of the sequence-based enabling condition.

6. The robot of claim 4, wherein the set of routines has a sensor-based enabling condition, the controller discontinuing execution of the predetermined sequence upon cessation of the sensor-based enabling condition.

7. The robot of claim 4, wherein routines of the set having a sequence-based enabling condition have a lower priority than routines of the set having only sensor-based enabling conditions.

8. The robot of claim 4, wherein the one or more working routines comprise a docking routine and a navigation routine, the docking routine having a sensor-based enabling condition and a higher priority than the navigation routine having both a sensor-based enabling condition and a sequence-based enabling condition.

9. The robot of claim 4, wherein the one or more working routines comprise, a coverage routine and a navigation routine, the coverage routine having a sensor-based enabling condition and a lower priority than a navigation routine having both a sensor-based enabling condition and a sequence-based enabling condition.

10. The robot of claim 4, wherein at least one routine of the set comprises a sequenced-reactive routine having at least one sensor-based enabling condition and at least one sequence-based enabling condition, wherein at least partial execution of another routine of the set satisfies the sequence based enabling condition of the sequenced-reactive routine, and wherein the controller aborts the sequenced-reactive routine upon cessation of at least one of the enabling conditions of the sequenced-reactive routine.

11. The robot of claim 10, wherein at least two sequenced-reactive routines share at least one sensor-based enabling condition or sequence-based enabling condition, thereby allowing multiple potential entry points into the predetermined sequence upon satisfying the shared enabling condition.

12. The robot of claim 4, wherein the specified sensor detection comprises at least one of detection of a navigation pose and detection of an obstacle proximate the robot.

13. The robot of claim 4, wherein the memory further stores a sequence planner executable by the controller to monitor a current state of the predetermined sequence and to set a sequence-based enabling condition according to a current sequence state of the robot upon detection of a sequence initiating event.

14. The robot of claim 13, wherein the sequence planner is configured to set a sequence-based enabling condition according to a current location state of a monitored topological state map.

15. The robot of claim 13, wherein the sequence planner is configured to allow execution of a routine by the controller based on a monitored state of a finite state machine stored in the memory.

16. The robot of claim 13, wherein the sequence planner controls execution, via the controller, of at least one goal oriented sequence of routines, the routines of the goal oriented sequence having a sequence-state-based enabling condition set by the sequence planner according to a current sequence state of the robot upon detection of a goal oriented sequence initiating event.

17. The robot of claim 4, wherein at least one of the escape routines has only a sensor-based enabling condition, and wherein the priority assigned to the at least one escape routine is higher than the priorities of all routines having sequence-based enabling conditions.

18. A method of navigating a robot, the method comprising: monitoring, for each routine in a set of routines, whether one or more enabling conditions assigned to the respective routine have been satisfied, the set of routines comprising a plurality of working routines and a plurality of escape routines, wherein each routine of the set is assigned a priority in addition to the enabling conditions, each of the plurality of escape routines having a higher priority than each of the plurality of working routines;identifying, from among the set of routines, a subset of routines for which all assigned enabling conditions have been satisfied, the subset of routines comprises at least one working routine and at least one escape routine;selecting, from among the identified subset of routines, a highest priority routine;permitting control according to the routines based on priority by permitting control according to the selected highest priority routine; and;driving the robot to navigate according to a predetermined sequence of control according to the routines of the set as the enabling conditions of the routines are successively satisfied.

19. The method of claim 18, wherein the one or more working routines comprise a docking routine and a navigation routine, the docking routine having a sensor-based enabling condition and a higher priority than the navigation routine having both a sensor-based enabling condition and a sequence-based enabling condition.

20. The method of claim 18, wherein the one or more working routines comprise a coverage routine and a navigation routine, the coverage routine having a sensor-based enabling condition and a lower priority than the navigation routine having both a sensor-based enabling condition and a sequence-based enabling condition.

21. The method of claim 18, including aborting execution of a routine upon cessation of an assigned enabling condition.

22. The method of claim 18, wherein routines at the set having a sequence-based enabling condition have a lower priority than each of the one or more escape routines and at least one avoidance routine.

23. The method of claim 18, wherein monitoring, for each routine in a set of routines, whether one or more enabling conditions assigned to the respective routine have been satisfied comprises: monitoring whether one or more sensor-based enabling conditions are satisfied, each sensor-based enabling condition being satisfied upon a specified sensor detection; andmonitoring whether one or more sequence-based enabling conditions are satisfied, each sequence-based enabling condition being satisfied upon a successful execution of a specified preceding routine.

24. The method of claim 23, wherein permitting control according to the routines based on priority comprises: permitting control of a set of routines in a predetermined sequence, the set of routines having a sequence-based enabling condition from a parent set of routines,executing the set of routines in the predetermined sequence upon enablement, and resuming a predetermined sequence of the parent set of routines upon completion of the set of routines or cessation of the enabling condition.

25. The method of claim 23, wherein permitting control according to the routines based on priority comprises permitting control of a set of routines in a predetermined sequence, the set of routines having a sensor-based enabling condition; anddiscontinuing execution of the predetermined sequence upon cessation of the enabling condition.

26. The method of claim 23, wherein routines having a sequence-based enabling condition have a lower priority than routines having only one or more sensor-based enabling conditions.

27. The method of claim 23, wherein at least one routine comprises a sequenced-reactive routine having at least one sensor-based enabling condition and at least one sequence-based enabling condition, wherein at least partial execution of another routine satisfies the sequence-based enabling condition of the sequenced-reactive routine, and wherein the robot aborts the sequenced-reactive routine upon cessation of at least one of the enabling conditions of the sequenced-reactive routine.

28. The method of claim 27, wherein at least two sequenced-reactive routines share at least one sensor-based enabling condition or sequence-based enabling condition, thereby allowing multiple potential entry points into the predetermined sequence upon satisfying the shared enabling condition.

29. The method of claim 23, wherein the specified sensor detection comprises at least one of detection of a navigation pose and detection of an obstacle proximate the robot.

30. The method of claim 23, further comprising monitoring a current state of a predetermined sequence of routines, andsetting a sequence-based enabling condition according to a current sequence state of the robot upon detection of a sequence initiating event.

31. The method of claim 30, wherein permitting control according to the routines based on priority comprises executing at least one goal oriented sequence of routines, the routines of the goal oriented sequence having a sequence-state-based enabling condition set by a sequence planner according to a current sequence state of the robot upon detection of a goal oriented sequence initiating event.

32. The method of claim 23, wherein at least one of the escape routines has only a sensor-based enabling condition, and wherein the priority assigned to the at least one escape routine is higher than the priorities of all routines having sequence-based enabling conditions.

33. A robot comprising: a chassis;a wheeled drive supporting the chassis and operable to maneuver the robot;sensors responsive to an environment about the robot;a memory storing: instructions configured as a set of routines to cause the wheeled drive to move the robot according to a predetermined sequence based on enabling conditions that must be satisfied, wherein the set of routines comprises a plurality of working routines and a plurality of escape routines, each routine of the set is assigned a priority and one or more enabling conditions, andwherein at least one of the escape routines comprises: at least one sensor-based enabling condition satisfied upon a specified sensor detection,at least one of the working routines comprises at least one sequence-based enabling condition satisfied upon a completion of a specified routine in the predetermined sequence, andat least one other routine of the set comprises a sequenced-reactive routine having at least one sensor-based enabling condition and at least one sequence-based enabling condition, wherein at least partial execution of another routine satisfies the sequence-based enabling condition of the sequenced-reactive routine, and wherein the controller aborts the sequenced-reactive routine upon cessation of at least one of the enabling conditions of the sequenced-reactive routine; andinstructions to enable control of the wheeled drive according to the assigned priorities and enabling conditions, the instructions configured to control the wheeled drive according to a highest priority routine of the set for which all assigned enabling conditions have been satisfied; anda controller in communication with the wheeled drive, the sensors, and the memory, the controller executing the instructions to: identify, from among the set of routines, a subset of routines for which all assigned enabling conditions have been satisfied, the subset of routines comprising one or more working routines and one or more escape routines;select, from among the identified subset of routines, an escape routine having the highest priority of the subset of routines; andcontrol the wheeled drive according to the selected escape routine.

34. The robot of claim 33, wherein at least two sequenced-reactive routines share at least one sensor-based enabling condition or sequence-based enabling condition, thereby allowing multiple potential entry points into the predetermined sequence upon satisfying the shared enabling condition.

35. A robot comprising: a chassis;a wheeled drive supporting the chassis and operable to maneuver the robot;sensors responsive to an environment about the robot;a memory storing: instructions configured as a set of routines to cause the wheeled drive to move the robot according to a predetermined sequence based on enabling conditions that must be satisfied, wherein the set of routines comprises a plurality of working routines and a plurality of escape routines, and each routine of the set effects movement of the wheeled drive,wherein each routine of the set is assigned a priority and one or more enabling conditions, andwherein at least one of the escape routines comprises at least one sensor-based enabling condition satisfied upon a specified sensor detection and at least one of the working routines comprises at least one sequence-based enabling condition satisfied upon a completion of a specified routine in the predetermined sequence, andinstructions to enable control of the wheeled drive according to the assigned priorities and enabling conditions, the instructions configured to control the wheeled drive according to a highest priority routine of the set for which all assigned enabling conditions have been satisfied; anda controller in communication with the wheeled drive, the sensors, and the memory, the controller executing the instructions, as enabling conditions are satisfied, to: identify, from among the set of routines, a subset of routines for which all assigned enabling conditions have been satisfied, the subset of routines comprising one or more working routines and one or more escape routines;select, from among the identified subset of routines, an escape routine having the highest priority of the subset of routines; andcontrol the wheeled drive according to the selected escape routine.