Adaptive mapping with spatial summaries of sensor data

Abstract

A system and method for mapping parameter data acquired by a robot mapping system is disclosed. Parameter data characterizing the environment is collected while the robot localizes itself within the environment using landmarks. Parameter data is recorded in a plurality of local grids, i.e., sub-maps associated with the robot position and orientation when the data was collected. The robot is configured to generate new grids or reuse existing grids depending on the robot's current pose, the pose associated with other grids, and the uncertainty of these relative pose estimates. The pose estimates associated with the grids are updated over time as the robot refines its estimates of the locations of landmarks from which determines its pose in the environment. Occupancy maps or other global parameter maps may be generated by rendering local grids into a comprehensive map indicating the parameter data in a global reference frame extending the dimensions of the environment.

Description

BACKGROUND OF THE INVENTION

Field of the Invention

The invention relates to a technique for generating a map of an environment using a plurality of sub-maps. In particular, the invention relates to a system and method for combining sensor data into a plurality of sub-maps based upon the location of the sensor when the data was acquired and the certainty with its location was known.

Description of the Related Art

In the past few years, a substantial research effort has been devoted to the problem of Simultaneous Localization and Mapping (SLAM). The term “map” in the field of SLAM generally refers to a spatial arrangement of observed landmarks or features. If these landmarks correspond to obstacle locations (such as the measurements collected with a Laser Range Finder), then the “map” yields an occupancy map denoting the floor plan of the space in which the robot is operating. In other cases, in which the landmark information does not correspond to obstacle locations (such as the measurements taken with a camera), the “map” estimated with SLAM techniques is dissociated from the locations of obstacles (occupancy map). However, an occupancy map is required for the robot to properly make decisions and navigate the environment.

A number of SLAM techniques have been proposed for simultaneously estimating the poses (i.e. localization) and building the map. Some methods re-estimate past poses instead of only the latest pose as new information is collected, achieving an improvement in the estimate of the robot trajectory as the localization system is updated. Laser scans, for example, are collected as a mobile robot moves through an indoor environment. These scans are combined with odometry information to estimate the robot's trajectory to yield a map showing the floor plan of the building. As more information is collected, the accuracy of the map improves because the estimates of the past poses of the robot are improved. A disadvantage of this system is that all sensor readings and their associated poses must be stored to allow the sensor data to be re-processed when new information arrives. This results in storage requirements that grow linearly with time. There is therefore a need for a localization and mapping technique that efficiently creates an occupancy map using new information to improve accuracy of the map without the storage requirement growing linearly with time.

SUMMARY OF THE INVENTION

The invention in the preferred embodiment features a system and method for mapping parameter data acquired by a robot or other mapping system that travels through an environment. The method generally comprises: measuring parameters that characterize the environment while driving the robot through the environment; generating estimates of the current robot pose, mapping parameter data to a current grid associated with an anchor node until the estimated pose uncertainty between with the current pose and the prior anchor node exceeds a threshold. When the threshold is exceeded, the robot generates a new grid associated with a new anchor node to record parameter data. The robot repeatedly generates new grids associated with different anchor nodes for purpose of recording parameter data. The estimated positions of the anchor nodes are updated over time as the robot refines its estimates of the locations of landmarks from which it estimates its position in the environment. When. an occupancy map or other global parameter map is required, the robot merges local grids into a comprehensive map indicating the parameter data in a global reference frame.

In accordance with some embodiments of the invention, the robot may map new parameter data to a new local parameter grid or to a pre-existing parameter grid. Data is recorded to a pre-existing parameter grid if the uncertainty between the current robot pose estimate and the pose estimate associated with the pre-existing grid is below a predetermined threshold. By using pre-existing grids, the robot can limit the memory requirements necessary to map the environment without the memory requirements growing linearly in time.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention is illustrated by way of example and not limitation in the figures of the accompanying drawings, and in which:

FIG. 1 is a functional block diagram of a robotic system, in accordance with the preferred embodiment of the present invention;

FIG. 2A is a diagrammatic illustration of the of a mobile robotic system in a global reference frame, in accordance with the preferred embodiment of the present invention;

FIG. 2B is a diagrammatic illustration of a local grid at a location coinciding with an anchor node in the global reference frame, in accordance with the preferred embodiment of the present invention;

FIG. 3A is a robot trajectory showing nodes and corresponding sensor data, in accordance with the preferred embodiment of the present invention;

FIG. 3B is a robot trajectory showing an anchor node and summary of sensor data, in accordance with the preferred embodiment of the present invention;

FIG. 3C is a robot trajectory showing an anchor node and summary of sensor data, in accordance with the preferred embodiment of the present invention;

FIG. 4 is a flowchart showing process of summarizing sensor data, in accordance with the preferred embodiment of the present invention;

FIG. 5A is a robot trajectory showing nodes, in accordance with the preferred embodiment of the present invention;

FIG. 5B shows a plurality of local parameter grids, in accordance with the preferred embodiment of the present invention;

FIG. 6A is a robot trajectory showing nodes, in accordance with the preferred embodiment of the present invention;

FIG. 6B shows a plurality of local parameter grids, in accordance with the preferred embodiment of the present invention;

FIG. 7A is an occupancy map depicting clear spaces in the environment explored by the robotic system, in accordance with the preferred embodiment of the present invention;

FIG. 7B is an occupancy map depicting obstacles in the environment explored by the robotic system, in accordance with the preferred embodiment of the present invention; and

FIG. 8 is a flowchart of the process of concurrent localization and parameter mapping, in accordance with the preferred embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

Illustrated in FIG. 1 is a functional block diagram of a mobile robotic system configured to generate spatial summaries as described in more detail below. The robotic system 100 includes one or more sensors 110, a central processing unit 130, one or more databases for storing relevant data, and a drive mechanism 150 including drive wheels 152, for example. The one or more sensors 110 include one or more visual sensors 112, i.e., cameras, video cameras, imagers including CCD imagers, CMOS imagers, and infrared imagers, for example, for acquiring images of the environment in which the robot is roving. The set of sensors in the preferred embodiment also includes one or more wheel odometers 158 for measuring the rotation of the wheels of the drive system. The set of sensors may further include one or more bump sensors 118 for generating a signal indicating the presence of an obstacle in the path of the mobile robot.

Data from the sensors 112, 114 may undergo preprocessing at processing unit 116. For example, the processing unit 116 may extract visual features from the image data for purposes of recognizing known landmarks, and process odometry data to convert wheel encoder signals or other odometry data to distance and rotation estimates. In some embodiments, odometry data may be used to detect and compensate for situations in which the drive wheels slip due to wet, slick, or carpeted surfaces. Data from the bump sensor 118 may undergo preprocessing at the processing unit 120 to determine when the robot encounters and obstacle as well as the position of the obstacle with respect to the robot path.

In other embodiments, the set of sensors 110 includes range finders, including laser, infrared (IR), and acoustic range finders; proximity sensors including lateral proximity sensors for determining lateral distance to objects in the environment; drop sensors for detecting staircases and other locations that are unsuitable for travel by the robot; and floor surface sensors, including sensors for measuring dirt concentration, slippage, and soil characteristics.

The mobile robot system 100 further includes at least one processor 130 configured to perform localization, generate maps of properties characterizing the environment in which the robot is operating, and navigate through the environment. In the preferred embodiment, the localization module 132 determines the location of landmarks as well as the mobile robot with visual and odometry data using a technique called Simultaneous Localization and Mapping (SLAM) 134 taught in U.S. Pat. No. 7,135,992 hereby incorporated by reference herein. Using this technique, the robotic system explores its environment, takes numerous images of its environment, makes a map depicting landmarks in the environment, and estimates the location of the robot relative to those landmarks. In the preferred embodiment, landmarks are visually identified using visual features from the image data are extracted and matched using a Scale Invariant Feature Transform (SIFT), Speeded Up Robust Features (SURF), Gradient Location and Orientation Histogram (GLOH), Binary Robust Independent Elementary Features (BRIEF), or other type of visual feature known to those skilled in the art. The visual landmarks—along with estimates of the robot position and orientation (pose) of the robot when the image was taken—are stored in the landmark database 142.

The parameter mapping module 136 is configured to generate a plurality of sub-maps or grids comprising local parameters and build global parameter maps based on those grids. In particular, the module 136 builds grids that depict the properties of the environment in proximity to associated anchor nodes, i.e, reference points fixed in their respective local reference frames. Estimates of the locations of the anchor nodes within the global reference frame are continually updated as the SLAM module 134 refines the localization map characterizing the environment. In the preferred embodiment, the parameters being mapped by the mapping module 136 include obstacles and clear spaces through which the robot system is free to navigate, as is explained in more detail below. Each of the anchor nodes is stored in node database 144 and the associated grid stored in the grid database 146. In the preferred embodiment, the mapping module includes an uncertainty tracking module 138 for measuring the uncertainty associated with the anchor nodes' localization estimate which is stored together with the anchor nodes' coordinates and heading in the global reference frame.

The processor 130 in the preferred embodiment further includes a navigation module 140 configured to generate signals that control the movement of the robot. For example, the navigation module can provide control signals to instruct the robot to move forward, to stop, to move backward, to turn, to rotate about a vertical axis. If the mobile robot system is an autonomous or semi-autonomous robot, the navigation module 140 can also perform path planning to efficiently guide the robot system to a desired destination and/or to achieve a desired goal. In accordance with the preferred embodiment, path planning is based on a parameter map that is generated from a plurality of parameter grids using current estimates of the poses of the anchors nodes corresponding to those grids.

The robot system 100 further includes a drive mechanism 150 for moving the robot around its environment, which may be indoors, outdoors, or a combination thereof. In the preferred embodiment, the drive mechanism includes two or more wheels drive wheels 152 powers by a motor 154 and battery pack 156, for example. In addition to, or instead of, the robot system may also incorporate other forms of locomotion including tracks, rollers, propellers, legs, and the like, to move around. The drive system may further include one or more optical wheel encoders 158, for example, for measuring the wheel rotation and estimating the distance traveled by the robot system. In addition, the difference in the rotation of opposing wheels can indicate changes in heading.

With wheel encoders 158 or other type of dead reckoning, the robot system can compute course and distance traveled from a previous position and orientation (pose) and use this information to estimate a current (pose). While relatively accurate over relatively short distances, dead reckoning sensing is prone to drift over time. Other forms of dead reckoning can include a pedometer (for walking robots), measurements from an inertial measurement unit, optical sensors such as those used in optical mouse devices, and the like.

In the preferred embodiment, the robot system 210 tracks its current location, path, or combination thereof with respect to a global reference frame represented by Cartesian (x-y) coordinates 250, as shown in FIG. 2. It will be understood that other coordinate systems, such as polar coordinates, can also be used. With respect to FIG. 2, a horizontal axis 252 corresponds to the x-axis and a vertical axis 254 corresponds to the y-axis. The origin 256 of the coordinate system may coincide with the robot's starting position, position or a prior anchor node, or other arbitrary location. The pose, including position and orientation, of the robotic system may be recorded in terms of the Cartesian coordinates and angle theta, θ.

By contrast, a grid in the preferred embodiment includes a map of local parameter data located relative to an anchor node in a local reference frame. As shown in FIG. 2B, properties of the environment in proximity to the anchor node are mapped to the grid 260 relative to the position of the anchor node A1. The grid 260 is therefore a local map describing the environment in the region around the anchor node. In the preferred embodiment, each grid includes a two dimensional Cartesian representation depicting the locations of obstacles (black cells) detect by the bump sensor 118 and open spaces (white cells) traversed by the robot (not to scale). In the preferred embodiment, an axis of the grid's Cartesian coordinate system coincides with robot's orientation anchor node, θ, which is generally different than the orientation of the x-axis 252 and y-axis 254 in the global reference frame. With respect to the global reference frame, an anchor node is typically a point along the path of the robot while navigating through the environment.

Although the grids in the preferred embodiment are shown as two dimensional (2D) Cartesian sub-maps, the grids may effectively record local parameter data using other reference systems including spherical and cylindrical coordinates systems for example. The parameter data is represented with pixels in a Cartesian coordinate system in the preferred embodiment. In alternative embodiments, grids may represent local parameter data as (1) pixels in a cylindrical coordinate system, (2) polygons with an arbitrary number of sides, or (3) other arbitrary shape, for example.

Referring to FIG. 3A, the robotic system 100 in the exemplary embodiment is configured to traverses a path through an environment. The path may be predetermined by the navigation module, determined ad hoc, or manually determined by a human driver or navigator, for example. While traversing the path, the localization module acquires image data with which it generates new landmarks and recognizes known landmarks for purposes of mapping the environment and locating the robotic system within the environment. The landmark information, in combination with the odometry information, enables the robotic system to make accurate estimates of the robot's location in the environment.

The robotic system generates a map of one or more parameters of interest in parallel with the location determination. In particular, the parameter mapping module senses properties of the environment and generates a parameter map depicting those properties. Referring to FIG. 3A, the mapping process begins by taking measurements of these properties and various locations or poses in the environment. The robot poses are represented as circles N1-N8 and the parameters observed at each of the respective poses is poses are represented as squares A-H. As one skilled in the art will appreciate, the parameter data would generally grow linearly in time as the robot continues to collect measurements. To limit the parameter to a manageable level, the robotic system in the present invention generates spatial summaries that effectively summarize parameter data is specific geographic locations. Referring to FIG. 3B, the robot system is configured to combine parameter readings for different poses if the relative uncertainty between those poses is small. For example, if Pose 1 and Post 2 in FIG. 3A have a relative pose transformation with low uncertainty, Sensor Data B and C can be combined into one summary corresponding to Pose A1 shown in FIG. 3B. The pose associated with the summary of Sensor Data Band C is tied to one root poses referred to herein as an anchor node. The pose selected to be the anchor node may be the pose associated with Pose 2, Pose 3, or a new pose created from the combination of the Pose 2 and 3.

Successive poses, like Pose 2 and Pose 3, generally have a relatively low relative uncertainty (due to the accuracy of the dead reckoning sensors) and may, therefore be combined into a single summary in many cases. As the localization information generated by the location module improves over time, the uncertainty of the relative pose between anchor nodes of two summaries will decrease. When the relative pose between two anchor nodes becomes sufficiently certain—the relative uncertainty drops below a threshold—the summaries associated with multiple nodes may be combined into a single summary that is then associated with a single anchor node. As shown in FIG. 3C, the summary of Sensor Data B and C is combined with the summary of Sensor Data H and G to create a single new summary associated with the anchor node A1. Since the new summary effectively summarizes the sensor data in the region of Anchor A1, the summary including Sensor Data B, C, G, and H is referred to herein as a spatial summary. As illustrated in FIG. 4, multiple pairs of anchor node poses are compared. In the extreme, the pose for each anchor is compared to the pose for every other anchor node. If the uncertainty associated with the relative pose between the anchor nodes is below a threshold, the decision block is answered in the affirmative and the summaries (comprised of sensor data) for the anchor nodes are combined into a single summary associated with a single anchor node. If, however, the uncertainty exceeds the threshold, the pair of anchor nodes is not combined and new sensor data added to the grid associated with the current node. The “uncertainty” of relative poses between anchor nodes is, in the preferred embodiment, the sum of the diagonal elements of the covariance matrix of the relative pose estimate. In other embodiments, the method of measuring relative uncertainty includes generating a Mahalanobis distance, or like uncertainty estimating metric.

As described above, the parameter data from a plurality of grids may be merged in a single summary associated with a single anchor nodes based on the relative pose uncertainty. Other criteria may also be used when determining whether to combine grids. These criteria may include, but are not limited to: (a) whether the summary reduces the memory requirements, i.e., whether the number of anchor nodes and grids data is reduced; (b) whether the summary improves performance, i.e., whether the summary reduces the time needed to compute a complete parameter map; (c) whether the map quality improves, i.e., whether merging or eliminating relatively “old” and outdated maps while retaining relatively “newer” maps improves the accuracy of the parameter map; or (d) any combination thereof

Illustrated in FIG. 5A is a robot path 500 and a plurality of corresponding nodes 510, and illustrated in FIG. 5B are the anchor nodes and associated grids that summarize the sensor data for the nodes 510 shown in FIG. 5A. Referring to FIG. 5A, the robot system collects sensor data while traverse a trajectory 500 through the environment. The sensor data, including obstacles, for example are associated with the pose of the robot at the time the sensor data was taken. Due to the volume of this data, however, the robotic system summarizes this data in the manner illustrated in FIG. 5B. Anchor nodes A1-A4 are shown in FIG. 5B as circles and the grids 520-523 shown as rectangles. In the preferred embodiment, the sensor data includes bump sensor data that indicates the presence of obstacles. Each grid, thus, depicts the locations of areas that are clear to traverse (shown as white cells) as well as obstacles or occupied areas (shown as black cells) in proximity to their respective anchor node.

In accordance with the preferred embodiment, the parameter mapping module 136 identifies nodes having a relative pose uncertainty below a threshold, combines the sensor data for these poses into a single grid, and associates the grid with a single anchor node. The parameter data from grids 520-523, for example, can be combined by overlaying the respective grids 520-523 as shown by the superposition 530 of grids. As one skilled in the art will appreciate, the plurality of grids may overlap in physical extent, possess different orientations in their respective local reference frames, and be of different sizes. Thereafter, data from the superposition 530 of grids may be combined into a single spatial summary associated with a new anchor node, for example. In the alternative, the superposition of spatial summaries may be used to build a global parameter map used to, for example, plan a new path for the robot through the environment. Exemplary parameter maps are shown and discussed in reference to FIGS. 7A and 7B.

Like FIG. 5A-5B, FIG. 6A-6B illustrates a robot path with corresponding nodes and anchor nodes with associated grids. As shown in FIG. 6A, the trajectory of the mobile robot has circled back on itself. In doing so, the robot traverses an area that it previously traversed earlier in its trajectory. As illustrated in FIG. 6B, by looping back, the robot is able to collect additional sensor data that can be used to update one or more previous grids and even modify sensor data used to populate the old version of the same grid. If the current pose of the robotic system is known with sufficient certainty relative to a prior pose, the anchor node associated with the prior pose is retrieved and the new sensor mapped to the grid associated with the prior anchor node.

For example, cells 520, 521 in the grid associated with anchor node A1 and A2 show occupied areas (or unsearched areas) in FIG. 5B. In FIG. 6B, the same cells 650, 652 in corresponding grids 620, 621 for anchor A1 and A2 were updated to show those cells as “clear areas” after the robot traverses the same area a second time. Similarly, new parameter data from sensors 110 is used to introduce new cells 654 to grid 523 in FIG. 5B to create the updated and expanded grid 623 in FIG. 6B. In both examples above, new sensor data collected while looping back is added to a prior grid because the uncertainty associated with the initial pose and later pose was below the acceptable threshold. In doing so, the mapping module 136 effectively updates existing grids with new information without creating new anchor nodes or grids. The present invention, therefore, effectively enables the parameter map to be continually updated with new sensor data without the storage requirements for the sensor data growing linearly with time.

At any point in time, the grids may be combined to generate a complete parameter map of the entire environment or a portion of the environment for purposes of path planning, for example. A representative parameter map is shown in FIGS. 7A and 7B. In the preferred embodiment, the plurality of grids depicting the presences of obstacles are combined to form an occupancy map of “clear” areas (i.e., open areas free of obstacles) in FIG. 7A and an occupancy map of “obstacles” (e.g., walls that bound the open areas) in FIG. 7B. Grids—also known as summaries when merged—may be combined by overlaying the grids at their respective locations in the global reference frame. The location of each individual grid is defined by the most current estimate of the position and orientation of the respective anchor point. The position and pose of each anchor node, in turn, is regularly updated within the global reference frame as new SLAM data is received and the uncertainties associated with the pose estimates is reduced. The occupancy map shown in FIGS. 7A and 7B are rendered in two dimensions (2D). In other embodiments, the occupancy map or other parameter map may be rendered in three dimensions (3D) if the sensor data and corresponding grids include elevational information, for example.

Illustrated in FIG. 8 is a flow chart showing the method of localization and parameter mapping, in accordance with the preferred embodiment of the present invention. In the preferred embodiment, the location and parameter mapping occur concurrently or substantially concurrently while the robotic system navigates 802 through the environment. With respect to localization, the robotic system repeatedly acquires images of the environment with which it identifies 804 landmarks. As the robot traverses the environment, it generally acquires multiple images or other measurements of each landmark which enables it to determine 806 the locations of the landmarks in two dimension (2D) or three dimensional (3D) space. As the map of landmarks is constructed and refined, the robot is able to make increasingly accurate estimates of its current pose 808 as well as the pose associated with each of the anchor nodes 810. The localization system may update the estimated locations of the anchor nodes to generate an occupancy map, for example, in the global reference frame. If an occupancy map is required for path planning for example, the decision block 812 is answered in the affirmative and the updated estimates of the locations of the anchor nodes used to superimpose the associated grids and render 814 the grids into a cohesive map as shown in FIGS. 7A and 7B.

While the robotic system navigates 802 through the environment, it measures 816 local parameters using on-board sensors including the bump sensor. Using the estimate of the current pose, the parameter mapping module searches for and identifies 818 an existing anchor node having the lowest relative pose uncertainty with respect to the current node. The identified node may be the preceding node in the robot path, or a prior node that is closest in distance to the current node. If the relative pose uncertainty between the current node and a prior node is below a predetermined threshold, the decision block 820 is answered in the affirmative. In this case, the grid associated with the prior anchor node is selected 822 to be the current grid and incoming sensor data mapped 826 to this current grid. The uncertainty is determined from the covariance matrix describing the positional uncertainties associated with the localization using the visual SLAM module and odometry sensors, for example. If, however, the uncertainty exceeds the predetermined threshold, the decision block 820 is answered in the negative. In this case, a new anchor node is generated 824 and the incoming sensor data mapped 826 to a new grid associated with the new anchor node. The process of mapping 826 incoming parameter data continues while the uncertainty remains sufficiently low. Over relatively short distances, dead reckoning measurements, such as those obtained from odometry readings, can be quite accurate. As such, the uncertainty remains low and incoming sensor data generally used to populate the current parameter. New nodes tend to be generated after the robot has traveled some distance in a previously unexplored area. New anchor nodes 830 are recorded in the node database 144 and new and updated grids 828 recorded in the grid database 146.

On occasion, the parameter data from a plurality of local grids is merged 832 into one or more spatial summaries. As discussed in detail in FIG. 4, grids may be combined into spatial summaries if the uncertainty associated with the relative pose between the respective anchor nodes is below a threshold. The mapping module 136 periodically attempts to generate spatial summaries in response to any of a number of events or conditions including: (1) elapse time; (2) space covered by the mobile robot or area mapped by the mobile robot; (3) grid memory limitation; (4) total number of grids or anchor nodes; or combination thereof. Moreover, the process of rending a plurality of grids into a global parameter map may be repeated as necessary based on the conditions stated above.

The robotic system of the present invention can be implemented in systems include hardware, software, firmware, or a combination thereof. Hardware can include one or more general purpose computers, microprocessors, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), and the like, as well as combinations thereof linked by networking systems, for example. Software may include computer-readable instructions for execution on various processors, computers, servers, or like circuit board or chips. The computer-readable instructions may be affixed in volatile or non-volatile memory including memory chips, hard drives, on compact discs, for example.

The present invention may also be implement in a plurality of platforms including a distributed platform including two or more network-enabled robots that cooperate with a remote central processing unit (CPU), for example, to collect landmark information from a relatively large environment. The CPU may include a personal computer, mobile phone, tablet computer, server, or like device that perform the computation of the processor 130. In some embodiments, the present invention is implemented with a fleet of robots that periodically exchange positioning information and parameter maps (either rendered a single map or as a collection of individual sub-maps) while traversing the environment so that each robot has information on all the parameters explored by other robots.

Although the description above contains many specifications, these should not be construed as limiting the scope of the invention but as merely providing illustrations of some of the presently preferred embodiments of this invention.

Therefore, the invention has been disclosed by way of example and not limitation, and reference should be made to the following claims to determine the scope of the present invention.

Claims

1. A method of mapping an environment, the method comprising: estimating a first current pose of a robot driving in an environment based on parameters measured by the robot, the robot having a visual sensor and the parameters including obstacles and clear spaces;defining a first local origin that represents an estimate of the first current pose, wherein the first local origin is one of a plurality of local origins;generating a first map of the measured parameters, wherein the measured parameters are mapped relative to the first pose;after driving a determined period of time, determining an estimate of a second current pose of the robot;determining an uncertainty between the estimate of the first current robot pose and the estimate of the second current pose of the robot; andresponsive to the uncertainty being greater than a first threshold, then:defining a second local origin that represents the estimate of the second current pose of the robot; andgenerating a second map of measured parameters mapped relative to the second current pose.

2. The method according to claim 1, wherein the visual sensor is configured to generate image data that includes features for recognizing a landmark in the environment.

3. The method according to claim 1, wherein the first map comprises a first sub-map and the second map comprises a second sub-map, andwherein the method further comprises generating a global map that includes data corresponding to the first sub-map and data corresponding to the second sub-map.

4. The method according to claim 1, wherein estimating the first current pose of the robot based on parameters measured by the robot is performed using sensor data that is generated by a plurality of sensors, the sensor data including data corresponding to the identification of obstacles and clear spaces that are included on the first map.

5. The method according to claim 1, wherein the first map and/or the second map include three-dimensional map data.

6. The method according to claim 1, after generating the first map and the second map, the method further comprising updating the first map and/or the second map based on defining the first local origin and/or the second local origin that are defined during robot travels occurring after the first map and the second map are generated.

7. The method according to claim 1, wherein the robot comprises a first robot, the method further comprising, after generating the first map and the second map, updating the first map and/or the second map based on map information that is generated by a second robot that is different from the first robot.

8. The method according to claim 1, further comprising merging, by the robot, system parameter data from a plurality of maps into a spatial summary.

9. The method according to claim 8, wherein the merging is performed responsive to one or more of: elapsed time, space covered by the robot or area mapped by the robot, a map memory limitation, or total number of maps or local origins.

10. The method according to claim 9, wherein the first local origin coincides with a starting position of the robot.

11. The method according to claim 1, wherein locations of obstacles detected by the robot within the environment comprise obstacles that are detected by a bump sensor.

12. The method according to claim 1, further comprising generating a spatial summary that is based on the measured parameters corresponding to the first map and the second map and that is associated with a local origin that is different from the first local origin and the second local origin.

13. The method according to claim 1, wherein the first local map comprises a map of local parameter data located relative to the first local origin, wherein the first local origin represents an estimate of the first current pose of the robot at a location.