Patent Grant
10954081
Parcel and other distributions centers may receive an arbitrary mix of items of various sizes, dimensions, shape, weight, rigidity, and/or other attributes, often in a cluttered arbitrary mix. Each item may have machine readable information, such as text and/or optically or otherwise encoded information, which can be machine read and used to route the item, e.g., via an automated sorting/routing system and/or processing. To read the information for a given item, in a typical approach the items are separated from one another via a process known as “singulation”.
Typically, singulation has been performed manually by human workers. A mixed of items arrives at a work station, e.g., via a chute or other conveyance, and each of a set of one or more human workers manually separates items and placed them in a defined space for a single item on a conveyor belt or the like. For each item, its destination (or at least next leg of transport) is determined by machine-reading information on the item, and the item is routed to a destination associated with the next leg, such as a bag, bin, container, or other receptacle and/or a delivery vehicle or staging area associated with the next leg.
Manual singulation processes are labor-intensive and can be inefficient. For example, a downstream human worker may have few locations on which to place singulated items, e.g., as a result of upstream workers filling many of the single item spots. Collective throughput may be suboptimal.
Use of robots to perform singulation is challenging due to the arrival of a cluttered mix of items at a work station, the dynamic flow of items at each station and overall, and the result that it may be difficult to identify, grasp, and separate (singulate) items using a robotic arm and end effector in an automated manner.
Various embodiments of the invention are disclosed in the following detailed description and the accompanying drawings.
The invention can be implemented in numerous ways, including as a process; an apparatus; a system; a composition of matter; a computer program product embodied on a computer readable storage medium; and/or a processor, such as a processor configured to execute instructions stored on and/or provided by a memory coupled to the processor. In this specification, these implementations, or any other form that the invention may take, may be referred to as techniques. In general, the order of the steps of disclosed processes may be altered within the scope of the invention. Unless stated otherwise, a component such as a processor or a memory described as being configured to perform a task may be implemented as a general component that is temporarily configured to perform the task at a given time or a specific component that is manufactured to perform the task. As used herein, the term ‘processor’ refers to one or more devices, circuits, and/or processing cores configured to process data, such as computer program instructions.
A detailed description of one or more embodiments of the invention is provided below along with accompanying figures that illustrate the principles of the invention. The invention is described in connection with such embodiments, but the invention is not limited to any embodiment. The scope of the invention is limited only by the claims and the invention encompasses numerous alternatives, modifications and equivalents. Numerous specific details are set forth in the following description in order to provide a thorough understanding of the invention. These details are provided for the purpose of example and the invention may be practiced according to the claims without some or all of these specific details. For the purpose of clarity, technical material that is known in the technical fields related to the invention has not been described in detail so that the invention is not unnecessarily obscured.
A robotic system to perform singulation and/or sortation is disclosed. In various embodiments, a robotic system includes a robotic arm and end effector used to pick items from a source pile/flow and place them on a segmented conveyor or similar conveyance to be sorted and routed for transport to a downstream (e.g., ultimate addressed/physical) destination. In some embodiments, picked items are placed singly into nearby bins or other receptacles. In some embodiments, multiple robots are coordinated to maximize collective throughput. In various embodiments, one or more robots may be employed at a singulation station. A system may include multiple stations. Human workers may be employed at one or more stations. The robotic system in various embodiments may be configured to invoke (request) the assistance of a human worker, e.g., by teleoperation of a robotic arm, manual task completion, etc., for example to handle an item the robot cannot handle by fully automated processing and/or an item the robot has dropped, etc.
Parcel carriers, postal services, delivery services, large retailers or distributors, and other enterprise and government entities that handle, transport, and deliver items to and from diverse locations typically receive large quantities of items from various source locations, each to be delivered to a corresponding one of a variety of destination locations.
Machines exist to handle, sort, and route items, but to use machine readers and sorting equipment items may need to be spaced from one another and/or in a certain orientation to be able to have a label or tag read by a machine. Such spacing and orientation may need to be achieved in the course of a process of “induction” of items into a sorting/routing facility, and may be performed in connection with a “sorting” or “sortation” process, for example a process by which items to be delivered to diverse locations are sorted by general destination (e.g., region, state, city, zip code, street, by street number order, etc.).
Machine readers, such as radio-frequency (RF) tag readers, optical code readers, etc., may need items to be spaced apart from one another, a process sometimes referred to as “singulation”, to be able to reliably read a tag or code and for the system to associate the resulting information with a specific item, such as an item in a specific location on a conveyor or other structure or instrumentality.
In a typical induction/sortation process in a parcel sorting operation, for example, individual parcels may be picked from bulk piles and placed onto a moving conveyor or tilt tray sortation system. For most facilities, induction of this type is entirely manual.
A typical, manual parcel induction/sortation process may include one or more of the following:
In a typical manual induction/sortation process, manually (or machine) fed chutes via which parcels of a variety of shapes and sizes arrive in bulk in various orientations; parcels may have different dimensions, shapes, rigidity, packaging, etc.; typically human workers take packages from a chute feeding a station at which each works and places them one by one on an open partitioned or otherwise defined segment of a conveyor; finally, many workers each at a station populate locations on one or more conveyors with singulated parcels, to facilitate downstream machine processing, such as reading the code or tag and taking automated sorting action based thereon, such as routing each parcel to a location within the facility that is associated with a destination to which the parcel is to be delivered. The location may involve further sorting (e.g., more destination-specific location within the facility) and/or packing/loading the parcel for further shipment (e.g., truck or aircraft to further destination where further sorting and delivery will occur, loading on a truck for local delivery, etc.).
In various embodiments, a robotic system comprising one or more robotic arms performs singulation/induction. In the example shown in
In the example shown, one or more of robotic arm 202, end effector 204, and conveyor 208 are operated in coordination by control computer 212. In various embodiments, control computer 212 includes a vision system used to discern individual items and each item's orientation based on image data provided by image sensors, including in this example 3D cameras 214 and 216. The vision system produces output used by the robotic system to determine strategies to grasp the individual items and place each in a corresponding available defined location for machine identification and sorting, such as a partitioned section of segmented conveyor 208.
In various embodiments, a robotic system as disclosed herein includes and/or does one or more of the following, e.g., by operation of a control computer such as control computer 212:
In various embodiments, an arbitrary mix of items to be singulated may include parcels, packages, and/or letters of a variety of shapes and sizes. Some may be standard packages one or more attributes of which may be known, others may be unknown. Image data is used, in various embodiments, to discern individual items (e.g., via image segmentation). The boundaries of partially occluded items may be estimated, e.g., by recognizing an item as a standard or known type and/or extending visible item boundaries to logical estimated extents (e.g., two edges extrapolated to meet at an occluded corner). In some embodiments, a degree of overlap (i.e., occlusion by other items) is estimated for each item, and the degree of overlap is taken into consideration in selecting a next item to attempt to grasp. For example, for each item a score may be computed to estimate the probability of grasp success, and in some embodiments the score is determined at least in part by the degree of overlap/occlusion by other items. Less occluded items may be more likely to be selected, for example, other considerations being equal.
In various embodiments, multiple 3D and/or other cameras may be used to generate image data. A 3D view of the scene may be generated, and/or in some embodiments a combination of cameras is used to look at the scene from different angles and the camera that is least occluded, e.g., with respect to a workspace and/or one or more specific items in the workspace, is selected and used to grasp and move one or more items.
The multiple cameras serve many purposes, in various embodiments. First they provide a richer full 3D view into the scene. Next they operate in cohesion to minimize the errors due to package shininess when light reflecting off a package and into a camera may disrupt its operation; in this case another camera at a different location provides a backup. In some embodiments, they can be selectively triggered by a predictive vision algorithm that determines which camera has the best viewing angle and/or lowest error rate for picking a particular package; as such each package has the optimal camera looking at it. In some embodiments, one or more cameras are mounted on an actuated base, of which the system can change the position and orientation to provide a more optimal perception (e.g., view) of a package.
Another purpose served by cameras is to detect any sort of unforeseen error in robot operation or any disruption to the environment. Cameras placed on the robot and on the environment have different error and accuracy profiles. The cameras on the robot can be more accurate since they are rigidly fixed to the robot but slower to use since using them requires the robot to slow down or stall. Cameras in the environment have a stable view and are effectively faster since the robot can multi-task and do something else while a camera is taking a photo. But if someone moves or shakes the camera stand, they would be out of sync with the robot and cause a lot of errors. Combining images from robot and non-robot cameras (occasionally or on a package miss) enables detecting if the robot is in sync with non-robot cameras, in which case the robot can take corrective action and be more robust. In some embodiments, a camera may not be mounted rigidly on a robotic arm, and in some such embodiments gyros and/or accelerometers on the cameras may be used to filter or compensate for the motion of the mounting base.
Referring further to
In various embodiments, control computer 212 coordinates operation of the four robotic arms 202, 236, 238, and 240 and associated end effectors, along with conveyor 208, to pick/place items from the chutes 206, 236, 238, and 240 to conveyor 208 in a manner that maximizes the collective throughput of the system.
While in the example shown in
In various embodiments, a scheduler coordinates operation of a plurality of robots, e.g., one or more robots working at each of a plurality of stations, to achieve desired throughput without conflict between robots, such as one robot placing an item in a location the scheduler has assigned to another robot.
In various embodiments, a robotic system as disclosed herein coordinates operation of multiple robots to one by one pick items from a source bin or chute and place them on an assigned location on a conveyor or other device to move items to the next stage of machine identification and/or sorting.
In some embodiments, multiple robots may pick from a same chute or other source receptacle. In the example shown in
While stationary robotic arms are shown in
At 322, the system attempts to grasp one or more items using the strategy determined at 320. For example, the end effector of a robotic arm may be moved to a position adjacent to the item(s) to be grasped, according to the determined strategy, and the end effector may be operated according to the determined strategy to attempt to grasp the item(s).
At 324 a determination is made as to whether the grasp was successful. For example, image data and/or force (weight), pressure, proximity, and/or other sensor data may be used to determine whether the item was grasped successfully. If so, at 326 the item(s) is/are moved to the conveyor. If not, processing returns to step 320, at which a new strategy to grasp the item is determined (if available).
In some embodiments, if after a prescribed and/or configured number of attempts the system fails to grasp an item, or if the system cannot determine a further strategy to grasp the item, the system moves on to identify and grasp another item, if available, and/or sends an alert to obtain assistance, such as from a human worker.
In the example shown, 3D scene visualization 400 comprises the output of a vision system as implemented in various embodiments. In this example, normal vectors to item surfaces are shown. Normal vectors are used in some embodiments to determine a strategy to grasp an item. For example, in some embodiments, the robot has a suction-type end effector and the normal vectors are used to determine an angle and/or orientation of the end effector to maximize the likelihood of a successful grasp using suction.
In some embodiments, the vision system is used to discern and distinguish items by object type. The object type may be indicated in visualization 400 by highlighting each object in a color corresponding to its type. In some embodiments, object type information is or may be used to determine and/or modify a grasp strategy for the item, such as by increasing suction pressure or grasp force, reducing movement speed, using a robot with a particular required end effector, weight capacity, etc.
In some embodiments, additional information not shown in
In the example shown, at 422 3D image data is received from one or more cameras, such as cameras 214, 216, 224, and/or 226 of
In various embodiments, the segmentation data determined at 424 is used to generate for each of a plurality of items a corresponding mask layer. The respective mask layers are used in various embodiments to generate and display a visual representation of the scene in which the respective items are highlighted. In various embodiments, each item is highlighted in a color corresponding to an object type and/or other attribute (weight, class, softness, deformability, unpickability—e.g., broken package, porous surfaces, etc.) of the item.
In the example shown, at 502 image data is received. Image data may be received from a plurality of cameras, including one or more 3D cameras. Image data may be merged to generate a 3D view of the workspace, such as a pile or flow of items in a chute or other receptacle. At 504, segmentation, object type identification, and normal vector computation are performed. In some embodiments, segmentation data is used to discern item boundaries and generate item-specific mask layers, as described above.
At 506, grasp strategies are determined for as many items as possible given the current view of the workspace. For each item, one or more grasp strategies may be determined, and for each strategy a probability that the strategy will result in a successful grasp of the item is determined. In some embodiments, a grasp strategy is determined based on item attributes, such as item type, size, estimated weight, etc. In some embodiments, determining a grasp strategy includes determining whether to attempt to pick up one item or multiple items, whether to use all end effector actuators (e.g., use just one or two or more suction cups or sets of cups) and/or selecting a combined grasp technique and corresponding speed from a set of computed grasp technique+speed combinations under consideration (e.g., 1 suction cup @ 50% speed vs. 2 suction cups 80% speed). For example, in some embodiments, a determination may be made based on package type to use one suction cup at 50% speed, such as to grasp a smaller item (can't use two suction cups) that may be heavy or difficult to grasp securely with one suction cup (move 50% speed instead of 80%, so as not to drop). A larger item that is also relatively light or easier to grasp securely with two suction cups, by contrast, may be grasped with two cups and moved more quickly.
In some embodiments, the grasp strategy may change if the object is moving as part of a flowing pile. For example, according to one strategy the robot may push down on the object harder to freeze it in place while the suction cups can be independently activated in the order in which they touch the package to ensure a tight and stable grasp. The robot will also match speed, based on visual or depth sensor feedback, with the flowing pile to ensure that the package doesn't slip by in the pile.
Another optimization implemented in some embodiments is to change grasp strategy based on double (or n−) item picking (also see next section) where a robot may adapt its strategy based on whether it's trying to reuse vision data while picking multiple items in a batch or sequentially. For example, a robot gripper with 4 suction cups can use 2 cups to pick up one object, and the others 2 to pick up a second object. In this scenario, the robot would lower its speed and would also approach the grasp for the second object in a manner that avoids collisions between the already held first object and the surrounding pile of objects.
At 508, the item-specific grasp strategies and probabilities determined at 506 are used to determine a plan to grasp, move, and place items according to the probabilities. For example, a plan to grasp the next n items in a specific order, each according to the grasp strategies and corresponding probabilities of success determined for that item and strategy, may be determined.
In some embodiments, changes in the location, arrangement, orientation, and/or flow of items resulting from the grasping and removal of earlier-grasped items is taken into consideration in formulating a plan, at 508, to grasp multiple items in succession.
In various embodiments, the process 500 of
Each of the robotic singulation station schedulers 524, 526, 528, and 530 is associated with a corresponding robotic singulation station and each controls and coordinates the operation of one or more robotic arms and associated end effectors to pick items from a corresponding chute or other item receptacle and place them singly on a segmented conveyor or similar structure. Each of the robotic singulation station schedulers 524, 526, 528, and 530 is associated with a corresponding set of one or more station sensors 532, 534, 536, and 538, respectively, and each uses the sensor data generated by its station's sensors to perform automated singulation at its robotic singulation station. In some embodiments, each implements and performs process 500 of
In various embodiments, each of the robotic singulation station schedulers 524, 526, 528, and 530 reports to global scheduler 522 one or more of image and/or other station sensor data; object identification, grasp strategy, and success probability data; pick/place plan information; and expected item singulation throughput information. Global schedule 522 is configured to use information received from the robotic singulation station schedulers 524, 526, 528, and 530—along with sensor data received from other sensors 540, such as cameras pointed at the segmented conveyor and/or other parts of the workspace not covered or covered well or completely by the station sensors—to coordinate work by the respective robotic singulation stations, each under the control of its station-specific scheduler 524, 526, 528, or 530, and to control the operation (e.g., speed) of the segmented conveyor via conveyor controller 542, so as to optimize (e.g., maximize) the collective singulation throughput of the system.
In various embodiments, the global scheduler 522 employs one or more techniques to optimize the use of a plurality of robots comprising the robotic singulation system to perform singulation, e.g., to maximize overall throughput. For example, if there are four robots in sequence, the lead (or other upstream) robot may be controlled to place packages in a manner that leaves open slots so that a downstream robot isn't waiting for an empty slot. This approach has impacts because downstream robots wait for some unknown/random amount of time because of package flow etc. As a result, a naive strategy (say lead robot places into every 4th slot empty) may not optimize collective throughput. Sometimes it might be better for the lead robot to put 2-3 packages into successive slots in sequence if its packages aren't flowing, but overall the system makes such decisions with awareness of state and flow at each station. In some embodiments, the optimal strategy for leaving open slots for downstream robots is based on an anticipated request for an open slot by the downstream robot (as a function of their package flow, for example). In some embodiments, information from the local station scheduler is used to anticipate the maximum throughput of each station and to control conveyor speeds and how many slots are left empty by upstream robots to ensure downstream robots have access to empty slots in proportion to the speed at which they are (currently) able to pick/place. In some embodiments, when the segmented conveyor is full due to some bottlenecks in the downstream sortation process, a robotic singulation system as disclosed herein may pre-singulate one or more packages, for example inside its corresponding chute or in a nearby staging area, while keeping tracking of the poses of each pre-singulated package. Once some empty spaces are available from the segmented conveyor, the system/station moves the pre-singulated packages onto the segmented conveyor, singly and in rapid succession, without additional vision processing time.
In some embodiments, the presence of humans working alongside robots has an impact on the placement and multi-robot coordination strategy since the robots or associated computer vision or other sensor system must now also watch what humans do and adapt the robot's placements in real-time. For example, if a human took over a conveyor belt slot that was scheduled to be used by a robot, the system must adjust its global and local schedules/plans accordingly. In another example, if a human disrupts a robot's picked package and causes it to register as not picked the system adapts to correct the error. Or, if a human corrects a robot's errors in picking (robot was commanded to put a package into slot A but accidentally placed it straddling across slot A and adjacent slot B; and human places it into slot B though the system memory says the package is in slot A, the system must observe the human's action and adjust downstream robot actions.
In various embodiments, the global scheduler 522 may cause a station to operate more slowly than its maximum possible throughput at a given time. For example, the global scheduler 522 may explicitly instruct the local station scheduler (e.g., 524, 526, 528, or 530) to slow down and/or may make fewer slots available to the local station, e.g., explicitly by assigning fewer slots to the station or indirectly, such as by allowing upstream stations to fill more slots.
In various embodiments, items arrive via a chute and/or conveyor. The chute/conveyor is fed, e.g., by humans, robots, and/or both, and items may arrive in clumps. With each addition to the flow at the input end, items may slide or flow down/along the chute or conveyor toward an end at which one or more robotic arms are provided to perform singulation, e.g., by grasping individual items and moving each one by one to a corresponding single item location on an output conveyor.
In various embodiments, images from one or more 3D or other cameras are received and processed. Flow of items down/through the chute/conveyor feeding a singulation station is modeled based on the image data. The model is used to predict a moment of relatively little or no movement, and at such a moment image data is used to identify and grasp an item. In some embodiments, the flow model is used to ensure the robotic arm is not in a position that would obscure the camera(s) at the moment of relatively little or no flow. In some embodiments, the system is configured to detect regions of relatively little or no flow within a broader flow. For example, flow may be analyzed within distinct, pre-define regions of a chute or other source conveyance or receptacle. The system may pick items for singulation from areas of little/stable and/or no flow and avoid for the time being regions in which less stable flow is occurring.
In some embodiments, the system identifies distinct areas of flow of items and guides the robot to pick the items accordingly from such areas. In some embodiments, the system predicts the motion of items on/through the chute or other source conveyance or receptacle, e.g., using a flow model computed using successive images from multiple cameras mounted over and around the chute. The flow model is used to predict a future position of one or more items as the item(s) flow(s) down the chute or other conveyance structure. For each time, an image is captured and/or grasp attempt at a time/location in which the item is expected to be based on the flow model. In some embodiments, real time segmentation is performed. For example, segmentation is performed within 30-40 milliseconds and/or some other rate that is as fast as the 3D camera frame rate. In some embodiments, real time segmentation results are used to track and/or model the flow of an individual item through the chute. A future position of an item is predicted based on its item-specific model/movement and a plan and strategy to grasp the item at the future location and time is determined and executed autonomously. In some embodiments, the position of the target item is updated continuously. In some embodiments, a trajectory of the robotic arm may be updated in real time, as the arm is in motion to grasp an item, based on a updated predicted future position of the item.
In some embodiments, if observed item flow is greater than a threshold speed, the system waits a configured amount of time (e.g., 100 milliseconds) and checks image data and/or flow rates again. The system may repeat and/or vary the length of such waits until a flow condition stable enough to have a high likelihood of successful grasp is observed. In some embodiments, different areas of the chute may have different speed thresholds which are allowed. For example if an area is far away from the actual spot where the robot is picking, a higher moving speed is tolerated in some embodiments, since unstable flow in that region may not be expected to disturb the operation being performed by the robot at that time. For example, a parcel which tumbles (“high speed”) at top of the chute may not be of concern if the robot is picking from the bottom of the chute. However, if the tumbling items were to get closer to the pick area, it may not be tolerable, since it can occlude other objects or move them around by hitting them. In some embodiments, the system detects flow by region and/or location relative to an anticipated pick area and tolerates less stable flow in area removed (sufficiently distant) from the area(s) from which the robot next expects to pick.
In the example shown in
Referring to
In the example shown in
In various embodiments, the system may take a photo and identify two (or more) objects to pick. The system picks and moves the first one; then, instead of doing a full scene re-compute to find a next package to pick, the system simply looks at whether the second package is disturbed. If not, the system picks it without doing a full recompute of the scene, which typically would save a lot of time. In various embodiments, the time savings is in one or more of sensor read latency, image acquisition time, system compute time, re-segmentation, masking, package pile ordering computations, and finally control and planning. If the second item is not where expected, the system does a full scene re-compute to find a next package to pick.
In some embodiments, a robot may use vision, depth, or other sensors to identify two graspable packages that are judged by the robot's control algorithm to be far enough apart so as not to be able to interfere with each other's picks (by destabilizing the pile and causing it to flow, or by hitting the other package etc.). Instead of repeating the sensor processing pipeline after picking the first package, the robot may directly proceed to pick the second package. That being said, statistically, there is a risk that the control algorithm's prediction may be incorrect or the pile may have moved due to some unforeseen event. In this scenario, the controller can pick in a more careful manner and use a model of predicted grasp suction cup activation (or other predicted sensory models, e.g., force, pressure, and other types of sensing modalities) to test whether the package being picked matches the earlier prediction (without having to re-do the vision computations).
In the example shown in
In the example shown in
Referring to
In various embodiments, more or fewer suction cups and/or more or fewer independently-actuated sets of one of more suction cup may be included in a given end effector.
When using a suction based end effector, such as end effector 802, pressing too hard to ensure suction cup engagement may damage fragile items and/or packaging. In addition, sensor readings may vary based on packaging type. For example, suction sensor readings when grasping corrugated boxes (air leaks from the grooves/paper) may be different from plastic poly bags. In some embodiments, if the package type is known, sensor readings are evaluated and/or adjusted accordingly. In some embodiments, if the sensor reading associated with successful grasp of a package of the determined type is not achieved on initial engagement, additional force and/or suction may be applied to achieve a better grasp. Such an approach increases the likelihood of a successful grasp and is more efficient than determining the grasp has failed and starting over.
In the example shown in
If it is determined at 908 that no further effort should be made at this time to grasp the item, the process advances to 912 at which it is determined whether there are other items available to be grasped (e.g., other items are present in the chute and the system has a grasp strategy available for one or more of them). If so, at 914 the system moves on to the next item and performs an iteration of step 902 and following steps with respect to that item. If there is no remaining item and/or no item for which a grasp strategy is available, then at 916 the system obtains assistance, e.g. from another robot, a human via teleoperation, and/or a human via manual intervention.
In various embodiments, the number and/or nature of further attempts to grasp an item, as in steps 908 and 910, may be determined by factors such as the item or item type, item and/or packaging characteristics determined by item type, item and/or packaging characteristics determined by probing the item with the robotic arm and end effector, items attributes as determined by a variety of sensors, etc.
In various embodiments, the robotic singulation station 1000 is controlled by a control computer configured to use a multi-view array of sensors, comprising cameras (or other sensors) 1010, 1012, and 1014, in this example, to scan address or other routing information locally, at the station 1000, at least in certain circumstances.
In various embodiments, sensors such as cameras (or other sensors) 1010, 1012, and 1014 are positioned to read routing information (e.g., text address, optical or other code, etc.) regardless of the orientation of the parcel as placed on the output conveyor by the robot. For example, if the parcel is placed label down a scanner across which parcel slides and/or is swiped reads the label and associates sorting/routing information with the corresponding location on the output conveyor.
The challenges with picking objects from piles (which may be flowing) and placing them on to a singulating conveyor is that the singulation conveyor needs to associate a package barcode with each package in a conveyor belt slot. Based on the package's orientation in the bulk pile from which the robot is picking, the robot might find a situation where the barcode actually faces down. In some embodiments, a robotic singulation system as disclosed herein may be used to flick or otherwise flip an item into a position in which the barcode or other label or routing information faces up. In some embodiments, the robot arm use its motions to flip the package by using controlled slipping due to gravity, if the end effector is pinch gripper (e.g., gripping at an end and letting the item begin to rotate via slippage/gravity before being release once the flipping motion has been initiated), or controlled suction release sequence on multiple suction cups and gravity to reorient a package (e.g., releasing suction on a cups at one end of the effector while still applying suction at the other end, to initiate rotation about an axis around which the item is to be flipped). However, flipping the package before placing it on the conveyor belt may damage a package, may not work, and/or may take a lot of time and could require another hand (e.g., another robot), which is expensive. In some embodiments, a system as disclosed herein may recognize an item is of a type that may require actions to ensure the label can be read. For example, a package in a polyethylene (“poly”) or other plastic or similar bag may need to be placed such that the packaging is flattened and/or smoothed out, for downstream scanners to be able to read the label. In some embodiments, a robotic arm may be used to smooth and/or flatten the packaging, e.g., after placement. In some embodiments, such an item may be dropped from a prescribed height, to aid in having the packaging become flattened enough to read. In various embodiments, the robot may flatten plastic bags by throwing them onto the chute or conveyor before picking and/or reading at the station, using actuated suction cups to stretch the bags after picking them up, performing bi-manual (two robotic arm) manipulation to un-wrinkle deformable items, etc. In some embodiments, a blower or other mechanism may be used to smooth the package after placement. The blower may be standalone/stationary mounted or integrate with/onto the robotic arm.
In some embodiments, a multi-axis barcode (or other) scanner or sensor is positioned downstream on the conveyor belt 1008, one that can scan barcodes on the top or bottom. However this only works if the conveyor belt has a transparent bottom or if there is a sequencing operation where the package is passed over a transparent (or open) slot with a barcode reader looking up through the slot.
In various embodiments, if a package cannot readily be placed on the conveyor 1008 with the label facing up, so it can be easily scanned by an overhead scanner, then an array of sensors at the singulation station, such as cameras (or other sensors) 1010, 1012, and 1014 in the example shown in
If the label is not on the top surface as grasped (1024), at 1028 the system uses one or more local sensors to attempt to find the label. If the label is found and is on a surface such that the robotic singulation system can place the item on the conveyor with the label rotated up, then at 1030 it is determined that a local scan is not necessary and the package is rotated and placed on the conveyor, at 1026, such that the label is up. If instead the label cannot be found or the package cannot be rotated to place it up, then it is determined at 1030 that the label should be scanned locally. In such a case, at 1032 the label is scanned locally, e.g., using a multi-axis sensor array such as the cameras 1010, 1012, and 1014 of
Downstream, the routing information determined by scanning the item's label locally, at the robotic singulation station, is used to sort/route the item to an intermediate or final destination determined based at least in part on the routing information.
In some embodiments, a robotic system includes one or more human singulation workers, e.g., at a last or other downstream station, as in the example shown in
In some embodiments, each singulation station includes room for one or more robots and one or more human workers, e.g., to work together at the same station. The location for a human worker provides sufficient space for the worker to perform singulation in an area in which the robotic arm will not extend. In various embodiments, the human worker may augment robotic arm throughput, fix misplacements, incorrect orientation (to scan label, for example), ensure conveyor slots not filled by an upstream robotic or other worker are filled, etc.
Considering that the singulation workspace is often limited the following situations are handled by the controller (e.g., control computer), in various embodiments:
Alternatively, it presents a pick decision on a software interface to a remote operator who guides the robot to make a right pick; the pick is a high level guidance provided by a human (not direct teleoperation) and is executed by the robot's picking logic above.
Alternatively, the robot nudges or pushes all the unpickable packages into a “return or unpickable item chute” that then routes them to a human handler.
In some embodiments, a single singulation station may have both human and robotic workers to pick and singulate items. The human may singulate items the human is trained to pick, such as items known to be difficult to pick and place via autonomous robotic operation. In some embodiments, the system monitors movement of the human worker and uses the robotic arm to pick, move, and place items via trajectories that avoid coming near the human worker. Safety protocols ensure the robot slows or stops its movement if the human gets too near.
At 1202, a placement error is detected. For example, one or more of images processed the vision system, force sensors on the robotic arm, pressure sensors detecting a loss of vacuum, etc. may be used to detect that an item was dropped before being placed. Or, image data may be processed to determine that an item was not placed on the intended slot on the conveyor, or that two items have been placed in the same slot, or that an item is placed in an orientation such that it will not be able to be scanned downstream, etc.
At 1204, the controller determines a plan for a downstream worker, such as another robotic arm and/or a human worker to fix the detected error. At 1206, the downstream worker is assigned to fix the error, and at 1208 to global and/or any affected local plans are updated, as needed, to reflect use of the downstream worker to fix the error. For example, the downstream worker may no longer be available, or available as soon, to perform a local task that the local scheduler may have assigned to it. In various embodiments, upon the resource (e.g., robotic arm) being assigned to fix an error detected upstream, the local scheduler will update its local plan and assigned tasks to incorporate the error correction task that has been assigned.
In another example, if a robot places two packages on to a slot that is supposed to hold one package, a downstream camera or other sensor can identify the error and share the information with a downstream robot. The downstream robot's control algorithm, in real-time, can deprioritize a pick, and instead pick the extra package and place it on an empty slot (or in its own pile to be picked and placed singly in a slot later). In some embodiments, a local barcode scanner may be used to scan the package, enabling the system to ensure the package barcode becomes associated with the slot in which it ultimately is placed.
In some embodiments, a quality monitoring system is provided to detect missed placements, dropped items, empty slots expected to contain an item, slots that contain more than one item, etc. In some embodiments, if there are more than one (or other required and/or expected) number of items in a slot of the segmented conveyor belt or other output conveyance the quality monitoring system checks to determine if some other slot is empty and takes corrective action to have the correct item moved from the slot having too many items to the empty slot. For example, a human worker or downstream robot may be tasked to pick the misplaced item from the slot having too many items and place it in the empty slot.
In various embodiments, techniques disclosed herein are used to provide a robotic singulation system capable of operating in most cases in a fully autonomous mode.
Although the foregoing embodiments have been described in some detail for purposes of clarity of understanding, the invention is not limited to the details provided. There are many alternative ways of implementing the invention. The disclosed embodiments are illustrative and not restrictive.
This application claims priority to U.S. Provisional Patent Application No. 62/926,165 entitled COORDINATING MULTIPLE ROBOTS TO MEET WORKFLOW AND AVOID CONFLICT filed Oct. 25, 2019 which is incorporated herein by reference for all purposes, and claims priority to U.S. Provisional Patent Application No. 62/993,579 entitled SINGULATION OF ARBITRARY MIXED ITEMS filed Mar. 23, 2020 which is incorporated herein by reference for all purposes.
| Number | Name | Date | Kind |
|---|---|---|---|
| 7313464 | Perreault | Dec 2007 | B1 |
| 20130083021 | Cohen | Apr 2013 | A1 |
| 20140257556 | Pandit | Sep 2014 | A1 |
| 20150225178 | Wargo | Aug 2015 | A1 |
| 20160167227 | Wellman | Jun 2016 | A1 |
| 20160221187 | Bradski | Aug 2016 | A1 |
| 20160354809 | Gruna | Dec 2016 | A1 |
| 20170080566 | Stubbs | Mar 2017 | A1 |
| 20180250823 | Shimodaira | Sep 2018 | A1 |
| 20180281200 | Rosenstein | Oct 2018 | A1 |
| 20190185267 | Mattern | Jun 2019 | A1 |
| 20200017317 | Yap | Jan 2020 | A1 |
| 20200086483 | Li | Mar 2020 | A1 |
| 20200130961 | Diankov | Apr 2020 | A1 |
| 20200311965 | Mizukami | Oct 2020 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 62926165 | Oct 2019 | US | |
| 62993579 | Mar 2020 | US |
