Embodied Robot Obstacle Avoidance and Looking Back Control Method, Device and Embodied Robot System

Abstract

The present application relates to the technical field of embodied robots, and discloses an obstacle avoidance and look-back control method, device and embodied robot system for an embodied robot. The method includes: in the case of triggering a soft collision, determining a plurality of look-back points on the initial contour according to the identified initial contour information of the current obstacle and the position at the time of triggering the soft collision; controlling the embodied robot to walk along the initial contour of the current obstacle, and when reaching each look-back point, performing a preset look-back action and obtaining a corresponding look-back result; updating the contour information of the current obstacle based on the look-back results that meet the preset conditions. By means of the look-back points and the preset look-back actions, the present application can update the obstacle contour information in real time, thereby improving the obstacle avoidance smoothness of the embodied robot and avoiding collisions.

Description

Technical Field

[0001] The present application relates to the technical field of embodied robots, and in particular, to an obstacle avoidance and look-back control method, device, and embodied robot system for an embodied robot. Background Art

[0002] In a robot system configured with a single obstacle avoidance sensor (such as a monocular camera), since the sensor can only obtain obstacle information from a single perspective, there are significant technical limitations when it performs the edge following obstacle avoidance task, which are specifically manifested in the following three aspects: First, the system stability is insufficient. When the robot moves around an obstacle, its movement trajectory shows obvious discontinuity, resulting in frequent jitter of the movement posture and affecting the overall running smoothness. Second, the distance control accuracy is poor. Due to the lack of multi-dimensional perception ability, there are large fluctuations in maintaining the distance between the robot and the obstacle, and it is difficult to ensure a constant safe distance. Finally, there is a risk of collision in the system. In certain situations, limited by the field of view and information acquisition ability of the sensor, the robot may not be able to identify the obstacle in time, thus resulting in a collision accident. Summary of the Invention

[0003] In view of this, to solve at least one of the above problems, embodiments of the present application provide an obstacle avoidance and look-back control method, device, and embodied robot system for an embodied robot.

[0004] In a first aspect, embodiments of the present application provide an obstacle avoidance and look-back control method for an embodied robot, including:

[0005] In the case of triggering a soft collision, according to the initial contour information of the currently recognized obstacle and the position when the soft collision is triggered, determine a plurality of look-back points on the initial contour;

[0006] Control the embodied robot to walk along the initial contour of the current obstacle, and when reaching each look-back point, perform a preset look-back action and obtain a corresponding look-back result;

[0007] Update the contour information of the current obstacle based on the look-back results that meet the preset conditions.

[0008] In an optional implementation manner, the look-back result is the new contour information of the currently recognized obstacle re-recognized after reaching a look-back point and performing the preset look-back action;

[0009] The updating the contour information of the current obstacle based on the look-back results that meet the preset conditions includes:

[0010] In the case that the deviation between the new contour information recognized at the current look-back point and the initial contour information exceeds a preset range, update the contour information of the current obstacle according to the new contour information.

[0011] In an alternative embodiment, it further includes: when the deviation between the newly recognized contour information and the initial contour information at the current look-back point does not exceed the preset range, controlling the embodied robot to go to the next look-back point and performing the preset look-back action to determine whether to perform the contour information update operation.

[0012] In an alternative embodiment, after each execution of the contour information update operation, the method further includes:

[0013] Based on the updated contour information of the current obstacle, re-determining a plurality of new look-back points;

[0014] Controlling the embodied robot to walk along the updated contour, and when reaching each new look-back point, performing the preset look-back action to determine whether to perform the contour information update operation again.

[0015] In an alternative embodiment, it further includes: during the process of controlling the embodied robot to walk along the current contour of the current obstacle, when not reaching the next look-back point, if a new obstacle is detected, and when no soft collision is triggered with the new obstacle, controlling the embodied robot to continue walking along the current contour of the current obstacle until reaching the next look-back point; wherein, the current contour is the initial contour or the updated contour;

[0016] In the case of triggering a soft collision with the new obstacle, controlling the embodied robot to switch to walking along the initial contour of the new obstacle, and when reaching each corresponding look-back point, performing the preset look-back action to determine whether to perform the contour information update operation of the new obstacle.

[0017] In an alternative embodiment, the determining a plurality of look-back points on the initial contour according to the initial contour information of the recognized current obstacle and the position when the soft collision is triggered includes:

[0018] Taking the connection line between the position coordinate point when the soft collision is triggered and the center point of the initial contour of the current obstacle as the reference line, and selecting at least one target line that forms a preset angle with the reference line;

[0019] Taking the non-soft collision intersection point of the reference line and the initial contour, and all intersection points of each target line and the initial contour as the selected plurality of look-back points.

[0020] In an alternative embodiment, before determining a plurality of look-back points on the initial contour according to the initial contour information of the recognized current obstacle and the position when the soft collision is triggered in the case of triggering a soft collision, it further includes:

[0021] Collect the visual information of the scene through the obstacle avoidance sensor;

[0022] Identify the obstacle type of the current obstacle according to the scene visual information and the preset obstacle types, and determine the initial contour information of the current obstacle based on the identified obstacle type; wherein, the initial contour information includes the polygon contour information of the current obstacle.

[0023] In an alternative embodiment, the obstacle avoidance sensor is in the direct front of the embodied robot; the execution of the preset look-back action includes:

[0024] Control the embodied robot to rotate by a specified angle and then retreat a preset distance, so that the current obstacle is within the viewing angle range of the obstacle avoidance sensor.

[0025] In an alternative embodiment, the determination process of triggering the soft collision includes:

[0026] Taking the position where the embodied robot is located as the starting point, construct a prediction area in the moving direction of the embodied robot;

[0027] If the prediction area contains points on the initial contour of the current obstacle or points inside the initial contour, trigger a soft collision.

[0028] In a second aspect, an embodiment of the present application provides an embodied robot obstacle avoidance look-back control device, including:

[0029] A determination module, configured to, when a soft collision is triggered, determine a plurality of look-back points on the initial contour according to the initial contour information of the identified current obstacle and the position when the soft collision is triggered;

[0030] An execution module, configured to control the embodied robot to walk along the initial contour of the current obstacle, and when reaching each look-back point, execute a preset look-back action and obtain a corresponding look-back result;

[0031] An update module, configured to update the contour information of the current obstacle based on the look-back results that meet the preset conditions.

[0032] In a third aspect, an embodiment of the present application provides an embodied robot system, including:

[0033] An execution mechanism for performing corresponding actions;

[0034] A memory, where a computer program is stored;

[0035] A processor, configured to execute the computer program to implement the embodied robot obstacle avoidance look-back control method in the foregoing embodiments.

[0036] The embodiments of the present application have the following beneficial effects: By selecting multiple look-back points on the initial contour of the obstacle and performing a preset look-back action at each look-back point, and then updating the contour information of the obstacle according to the qualified look-back results, the embodied robot can gradually complete the actual contour information of the obstacle during the look-back process. In this way, when encountering the same obstacle again, the obstacle avoidance process can be made smoother and jitter can be reduced, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for the embodiments. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0038] Figure 1 Shows a schematic structural diagram of an embodied robot system according to an embodiment of the present application;

[0039] Figure 2 Shows a first flowchart of an embodied robot obstacle avoidance look-back control method according to an embodiment of the present application;

[0040] Figure 3 Shows a second flowchart of an embodied robot obstacle avoidance look-back control method according to an embodiment of the present application;

[0041] Figure 4 Shows a third flowchart of an embodied robot obstacle avoidance look-back control method according to an embodiment of the present application;

[0042] Figure 5 Shows a schematic diagram of the construction of a prediction area according to an embodiment of the present application;

[0043] Figure 6 Shows a fourth flowchart of an embodied robot obstacle avoidance look-back control method according to an embodiment of the present application;

[0044] Figure 7 Shows a schematic diagram of the determination of look-back points according to an embodiment of the present application;

[0045] Figure 8 Shows a schematic diagram of the process of supplementing the obstacle information of an embodied robot according to an embodiment of the present application;

[0046] Figure 8 (a) of shows a schematic diagram of an embodied robot starting to walk along the initial contour of an obstacle after walking to a soft collision position according to an embodiment of the present application;

[0047] Figure 8(b) shows a schematic diagram of the embodied robot in an embodiment of the present application when walking along the initial contour of an obstacle and reaching one of the look-back points;

[0048] Figure 8 (c) shows a schematic diagram of the rotation by a specified angle action of the embodied robot in an embodiment of the present application when performing a preset look-back action;

[0049] Figure 8 (d) shows a schematic diagram of the backward movement by a preset distance action of the embodied robot in an embodiment of the present application when performing a preset look-back action;

[0050] Figure 8 (e) shows a schematic diagram of re-identifying the contour information of the current obstacle in an embodiment of the present application;

[0051] Figure 8 (f) shows a schematic diagram of re-determining the look-back point on the new contour information of the currently re-identified obstacle in an embodiment of the present application;

[0052] Figure 9 shows the fifth process schematic diagram of the obstacle avoidance and look-back control method of the embodied robot in an embodiment of the present application;

[0053] Figure 10 shows a schematic structural diagram of an obstacle avoidance and look-back control device of the embodied robot in an embodiment of the present application. Detailed implementation manners

[0054] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0055] Generally, the components of the embodiments of the present application described and shown in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application claimed, but merely represents the selected embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0056] As used below, the terms "comprising", "having" and their cognates that can be used in various embodiments of the present application are only intended to indicate specific features, numbers, steps, operations, elements, components or combinations of the foregoing items, and should not be construed as first excluding the existence of one or more other features, numbers, steps, operations, elements, components or combinations of the foregoing items or precluding the possibility of adding one or more features, numbers, steps, operations, elements, components or combinations of the foregoing items. In addition, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0057] Unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the various embodiments of the present application belong. The terms (such as those defined in a commonly used dictionary) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning unless clearly defined in the various embodiments of the present application.

[0058] The following will describe in detail some embodiments of the present application with reference to the accompanying drawings. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0059] For a robot system with a single obstacle avoidance sensor configuration, when performing an edge following obstacle avoidance task, it is found that the single obstacle avoidance sensor provided on the front side of the robot body may be easily blocked by the robot's own structure or external objects during the execution of the corresponding task, resulting in information loss. As a result, problems such as jitter, uneven distance from obstacles, and collisions with obstacles are likely to occur. For this reason, the present application proposes an embodied robot obstacle avoidance look-back control method, device, and embodied robot system. When a soft collision is triggered, the method determines a plurality of look-back points on the initial contour according to the recognized initial contour information of the current obstacle and the position at which the soft collision is triggered, and executes a preset look-back action when the embodied robot reaches each look-back point along the edge. Then, the contour information of the current obstacle is updated according to the look-back result to complete the contour information of the obstacle, thereby solving the problems during the edge following obstacle avoidance of the robot and making the entire obstacle avoidance process smoother and more stable.

[0060] Among them, the Embodied Robot in this application refers to a robot that has a physical body (entity) and realizes intelligent behavior through real-time perception, interaction, and action with the real environment. Its core idea stems from the theory of Embodied Intelligence, that is, intelligence not only depends on algorithms and data processing, but also needs to learn and evolve through the dynamic interaction between the body and the environment. Among them, the physical entity means that the Embodied Robot has a real body (such as a robotic arm, a mobile chassis, sensors, etc.), and can act in the physical world like a human or an animal (such as walking, grasping, obstacle avoidance, etc.). In addition, it can also sense the environment in real time through sensors such as vision, touch, hearing, and force perception, and adjust its behavior according to the feedback.

[0061] Figure 1 FIG. shows a schematic structural diagram of an Embodied Robot 10 according to an embodiment of the present application.

[0062] Exemplarily, the Embodied Robot 10 includes a processor 11, a memory 12, a sensing unit 13, and an actuator 14. Among them, the sensing unit 13 is used to detect the environmental information where the Embodied Robot is located; the actuator 14 is used to execute corresponding actions; the memory 12 stores a computer program, and the processor 11 runs the computer program, so that the Embodied Robot executes the Embodied Robot obstacle avoidance and look-back control method in the following embodiments.

[0063] Among them, the processor 11 can be an integrated circuit chip with signal processing capabilities. The processor 11 can be a general-purpose processor, including at least one of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a Network Processor (NP), a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc., and can implement or execute computer programs to implement the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application.

[0064] Among them, the memory 12 can be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), an Electric Erasable Programmable Read-Only Memory (EEPROM), etc. Among them, the memory 12 is used to store a computer program, and after receiving an execution instruction, the processor 11 can execute the computer program accordingly.

[0065] Among them, the sensing unit 13 can include detection sensors arranged on the body of the embodied robot 10. The detection sensors are mainly used to detect targets or obstacles on the traveling path of the embodied robot 10. For example, they can include visual sensors, lidars, etc. Among them, the visual sensor can include a camera, specifically, a conventional RGB (Red, Green, Blue) camera, a depth camera, a stereo camera, etc. For example, for a cleaning robot, its detection sensor can include a visual sensor. By collecting image data in the current environmental space through the visual sensor, information about obstacles in front in the environmental space can be obtained, and thus can be used for tasks such as target recognition and positioning. In some alternative embodiments, the detection sensors can also include infrared sensors, ultrasonic sensors, etc.

[0066] It should be understood that the above-mentioned embodied robot 10 can include, but is not limited to, a sweeping robot (also called a cleaning robot), a towing robot driven by a sweeper (i.e., a sweeping and towing integrated robot), a food delivery robot, an autonomously driven load-carrying robot, a companion robot, a service robot, etc. It can be understood that the existence forms of the various robots listed above are not limited. For example, it can be a wheeled robot, or a humanoid robot with two legs, or a multi-legged robot, etc.

[0067] Figure 2 A schematic flowchart of a method for obstacle avoidance and look-back control of an embodied robot according to an embodiment of the present application is shown. Exemplarily, the method for obstacle avoidance and look-back control of the embodied robot includes steps S210 - S230:

[0068] S210, in the case of triggering a soft collision, determine a plurality of look-back points on the initial contour according to the initial contour information of the currently identified obstacle and the position when the soft collision is triggered.

[0069] Among them, soft collision means that the embodied robot has a certain distance from the currently detected obstacle, but does not have a real collision with the obstacle. In this embodiment, by setting a soft collision mechanism, when the soft collision mechanism is triggered, it indicates approaching the current obstacle. At this time, the embodied robot can more intelligently adjust its own behavior during the obstacle avoidance process. For example, the robot can switch to a mode of moving along the edge of the obstacle (i.e., "edge avoidance"). During the edge avoidance process, the robot can determine to execute a preset look-back action at the selected look-back point to further optimize the contour information of the obstacle.

[0070] In some embodiments, before step S210, there is also a process of obtaining the initial contour information of the current obstacle, and this process is as Figure 3 shown, including steps S310 - S320:

[0071] S310, collect scene visual information through an obstacle avoidance sensor.

[0072] Among them, the obstacle avoidance sensor is mainly used for the embodied robot to sense obstacles in the surrounding environment. Generally, it can be set at the front or side of the robot body, and it can be, but is not limited to, a monocular camera.

[0073] It can be understood that during the process of the embodied robot executing the target task, the obstacle avoidance sensor will collect scene visual information in real time or at a preset time interval. Among them, the scene visual information can be image visual information or point cloud visual information.

[0074] S320, according to the scene visual information and the preset obstacle type, identify the obstacle type of the current obstacle, and determine the initial contour information of the current obstacle based on the identified obstacle type.

[0075] In the controller of the embodied robot 10 of this embodiment, there is an obstacle recognition module. Among them, some obstacle types are pre-stored in the obstacle recognition module, such as a ball of thread, a base, a shoe, a toy, a humidifier, pet supplies, feces, a dust collection station, building blocks, a doll, etc., and the contour information of these pre-stored obstacles. In one embodiment, the contour information mainly refers to the polygon contour information of each obstacle, such as including the coordinate information of each vertex of the polygon corresponding to the obstacle, the information of the sides, etc.

[0076] Exemplarily, after the obstacle avoidance sensor collects the scene visual information, it transmits the scene visual information to the obstacle recognition module. The obstacle recognition module uses a deep learning model (such as a YOLO architecture model, an object detection algorithm, or a semantic segmentation algorithm) to analyze the obtained scene visual information, and determines whether there is an obstacle in the scene visual information that is the same as the pre-stored obstacle type. If so, it determines that there is an obstacle in the current scene visual information and determines the type of the obstacle.

[0077] After determining the type of the current obstacle, the obstacle recognition module will continue to determine the initial contour information of the current obstacle according to the recognized type. Among them, the initial contour information includes the polygon contour information of the current obstacle. When obtaining the initial contour information, the embodied robot can obtain the preset size of the obstacle (the preset contour size of the obstacle) by combining the width of the obstacle detection frame (the obstacle detection frame is used to identify the position and size of the detected obstacle in the scene image, usually represented by a rectangular frame) in the detected size of the obstacle, and then obtain the estimated area of the obstacle, and collect the contour information of the obstacle according to the estimated area.

[0078] In some embodiments, as Figure 4 shown, the process of determining the trigger of soft collision includes steps S410 - S440:

[0079] S410, starting from the position of the embodied robot, construct a prediction area in the moving direction of the embodied robot.

[0080] S420, determine whether the points on the initial contour of the current obstacle or the points inside the initial contour are included in the prediction area.

[0081] If the points on the initial contour of the current obstacle or the points inside the initial contour are included in the prediction area, then execute step S430 to determine that the soft collision is triggered. On the contrary, if the points on the initial contour of the current obstacle or the points inside the initial contour are not included in the prediction area, then execute step S440 to determine that the soft collision is not triggered, and control the embodied robot to continue walking.

[0082] Exemplarily, if the embodied robot recognizes an obstacle through the obstacle avoidance sensor, it will continue to detect whether the embodied robot triggers a soft collision during the movement. As Figure 5 shown ( Figure 5 in which R is the embodied robot and V is the contour of the obstacle), a sector area with a radius of a predetermined length and a central angle of a predetermined angle can be constructed in the moving direction of the embodied robot starting from the center point coordinates of the embodied robot, and then this sector area is used as the above-mentioned prediction area.

[0083] After determining the prediction area, since the coordinates of each point included in the prediction area are known, then it is determined whether the points on the initial contour of the current obstacle or the points within the initial contour are included in the prediction area. If included, it means that there is an intersection between the prediction area and the current obstacle, that is, the soft collision condition is satisfied, so it is determined that a soft collision is triggered. On the contrary, if the points on the initial contour of the current obstacle or the points within the initial contour are not included in the prediction area, it means that there is no intersection between the prediction area and the current obstacle, so it is determined that no soft collision is triggered, that is, it means that the distance between the embodied robot and the current obstacle is still within an acceptable range and has no impact on the current movement of the embodied robot. Furthermore, the embodied robot can continue to be controlled to move forward until it is detected that there is an intersection between the prediction area and the polygonal contour and its internal points of the obstacle, and then step S430 is executed.

[0084] Among them, the predetermined length in the above prediction area can be determined according to factors such as the movement speed of the embodied robot, the effective detection range of the obstacle avoidance sensor (the predetermined length should not exceed the effective detection range of the obstacle avoidance sensor, otherwise the obstacle cannot be detected), and the actual application scenario. For example, if the movement speed of the embodied robot is relatively fast, the predetermined length needs to be larger so as to have enough time to detect potential collisions and react in time. If the movement speed of the robot is relatively slow, the predetermined length can be appropriately shortened. If in a narrow space (such as a home environment), the predetermined length can be shorter. If in an open space (such as a warehouse or an office), the predetermined length can be longer.

[0085] Among them, the predetermined angle in the above prediction area generally does not exceed the field of view angle range of the obstacle avoidance sensor, and can be specifically determined according to factors such as the obstacle avoidance accuracy requirements of the embodied robot.

[0086] In the above way, it can be determined whether the embodied robot and the current obstacle will trigger a soft collision, and the intersection area between the prediction area of the embodied robot and the current obstacle when a soft collision is triggered. Furthermore, the point closest to the center of the embodied robot in the intersection area between the prediction area and the current obstacle can be selected as the position coordinate point when a soft collision is triggered, as Figure 5 the position of the "○" in is the position coordinate point when a soft collision is triggered.

[0087] Among them, the look-back point refers to the position point where the embodied robot re-identifies the obstacle contour during the process of walking along the obstacle contour. In this embodiment, by setting the look-back point, the embodied robot can gradually complete the contour information of the obstacle, enabling the embodied robot to better understand the contour of the obstacle, and further enabling the embodied robot to execute the target task smoothly and safely. In this embodiment, multiple look-back points also need to be determined on the initial contour according to the initial contour information of the currently identified obstacle and the position when a soft collision is triggered.

[0088] Exemplarily, as Figure 6As shown, the process of determining multiple look-back points on the initial contour according to the initial contour information of the currently recognized obstacle and the position when the soft collision is triggered includes steps S610 - S620:

[0089] S610, taking the line connecting the position coordinate point when the soft collision is triggered and the center point of the initial contour of the current obstacle as the reference line, and selecting at least one target line that forms a preset angle with the reference line.

[0090] S620, taking the non-soft collision intersection point of the reference line and the initial contour, and all the intersection points of each target line and the initial contour as the selected multiple look-back points.

[0091] Exemplarily, first determine the center point of the initial contour of the current obstacle. It can be understood that after determining the initial contour information of a polygonal shape and knowing the coordinates of each vertex of the initial contour, the position of the center point of the initial contour of the current obstacle can be calculated through a mathematical formula.

[0092] After determining the position of the center point of the initial contour of the current obstacle, as Figure 7 shown, connect the position coordinate point when the soft collision is triggered and the center point of the initial contour of the current obstacle as the reference line, that is, connect the coordinate point at the "○" position on the initial contour with the coordinate point at the " " position inside the initial contour to form a reference line A. Then select at least one target line that forms a preset angle with the reference line A, where the preset angle can be 90 degrees, 45 degrees, 60 degrees, 135 degrees, etc. It can be understood that there are two intersection points between the reference line and the initial contour, one of which is the position coordinate point when the soft collision occurs (referred to as the soft collision intersection point here), and the other is called the non-soft collision intersection point. In addition, the more target lines are selected, the more intersection points there are with the initial contour, and correspondingly, the more the number of look-back points.

[0093] In order to be able to complete the contour information of the obstacle with fewer look-back times and complete the edge walking of the entire obstacle faster in the subsequent process, preferably, in one implementation, a target line perpendicular to the reference line (i.e., 90 degrees) can be selected. As Figure 7 shown, the target line is line B. In this way, two intersection points are formed between the target line and the initial contour. Adding the non-soft collision intersection point of the reference line and the initial contour, these three intersection points are the determined look-back points, that is, the position of " Figure 7 " in . It can be understood that the above soft collision intersection point is not selected as a look-back point here because the contour information obtained by the embodied robot at this point is the initial contour information. By continuing to walk along the initial contour and observing the current obstacle from other perspectives, the actual contour information of the current obstacle can be completed.

[0094] S220, control the embodied robot to walk along the initial contour of the current obstacle. When reaching each look-back point, execute a preset look-back action and obtain the corresponding look-back result.

[0095] Among them, the look-back result is the new contour information of the currently recognized obstacle after reaching a look-back point and executing the preset look-back action.

[0096] Exemplarily, after determining the look-back point, as Figure 8 shown in (a) of Figure 8 , control the embodied robot to enter the process of walking along the obstacle contour. Specifically, it continues to walk from the current position. When reaching the soft collision position, control the embodied robot to walk along the initial contour of the current obstacle. During the process of walking along the initial contour of the current obstacle, continuously judge whether a look-back point is reached. If one of the look-back points is reached (as Figure 8 shown in (b) of Figure 8 (b) of is a schematic diagram of the embodied robot reaching one of the look-back points when walking along the initial contour of the current obstacle)), trigger the preset look-back action. In

[0097] ]>, the embodied robot is R, "In one implementation, executing the preset look-back action includes: controlling the embodied robot to rotate by a specified angle and then retreat by a preset distance so that the current obstacle is within the viewing range of the obstacle avoidance sensor.

[0098] Exemplarily, as Figure 8 shown in (b) of Figure 8 , since the embodied robot walks along the contour of the current obstacle and the obstacle avoidance sensor is in the front of the embodied robot, that is, the viewing range of the embodied robot is at the position in the front of the robot. As ​ shown in (b) of ​As shown in (d), control the embodied robot to retreat a preset distance. The retreat is mainly to bring the current obstacle within the visual range of the embodied robot so as to re-identify the contour of the obstacle subsequently.

[0099] After performing the preset look-back action at the current look-back point, as ​ shown in (e), the embodied robot will re-acquire the contour information of the current obstacle to obtain the new contour information of the current obstacle ( ​ V’ in (e)), and this new contour information is the look-back result of the current look-back point. Among them, the process of re-identifying the contour information of the current obstacle is the same as the above-mentioned recognition process of the obstacle contour information, which will not be elaborated here.

[0100] S230, update the contour information of the current obstacle based on the look-back result that meets the preset conditions.

[0101] Exemplarily, when the deviation between the new contour information recognized at the current look-back point and the initial contour information exceeds the preset range, update the contour information of the current obstacle according to the new contour information. It can be understood that if there is a deviation between the recognized new contour information and the initial contour information, and the deviation is large, that is, it exceeds the preset range, it is considered that the recognized initial contour information is inaccurate. In order to make the contour information of the current obstacle more accurate, the recognized new contour information is updated as the contour information of the current obstacle.

[0102] As an optional solution, when the deviation between the new contour information recognized at the current look-back point and the initial contour information does not exceed the preset range, that is, the difference between the recognized new contour information and the initial contour information is not large, control the embodied robot to continue walking along the initial contour of the current obstacle to the next look-back point, and after reaching the next look-back point, perform the preset look-back action to determine whether to perform the contour information update operation. It can be understood that every time a look-back point is reached, a preset look-back action will be performed, and the contour information of the current obstacle will be re-recognized, and then it is judged whether the contour information of the current obstacle needs to be updated.

[0103] In some embodiments, after each execution of the contour information update operation, as ​ shown, the method further includes steps S910 - S920:

[0104] S910, re-determine a plurality of new look-back points based on the updated contour information of the current obstacle.

[0105] Since the contour information of the current obstacle has been updated, if the preset look-back action is still performed at the position of the previous look-back point, it may lead to incorrect or untimely recognition of the contour of the current obstacle. Therefore, after the contour information of the current obstacle is updated to the new contour information, it is necessary to re-determine the new look-back points (re-determining the new look-back points is as​ as shown in (f) thereof. The process of determining the new look-back point is the same as the process of determining the look-back point described above, and will not be elaborated here.

[0106] After updating the contour information of the current obstacle and determining the new look-back point, step S920 is executed to control the embodied robot to walk along the updated contour. When each new look-back point is reached, a preset look-back action is executed to determine whether to perform the contour information update operation again.

[0107] In this embodiment, by executing a preset look-back action every time a look-back point is reached and re-identifying the contour information of the obstacle once, the contour of the obstacle can be better verified and complemented, so that the finally determined contour information of the obstacle is accurate. In addition, by re-identifying the contour information of the obstacle every time a look-back point is reached and updating the contour information of the obstacle when the preset conditions are met, it can be ensured that the embodied robot can safely walk along the contour of the obstacle and avoid collisions.

[0108] As an optional solution, the obstacle avoidance look-back control method for the embodied robot further includes:

[0109] During the process of controlling the embodied robot to walk along the current contour of the current obstacle, it will be continuously judged whether there is a new obstacle. If a new obstacle is detected before reaching the next look-back point, it is judged whether the embodied robot has a soft collision with the new obstacle. In the case of no soft collision with the new obstacle, the embodied robot is controlled to continue walking along the current contour of the current obstacle until the next look-back point is reached; wherein, the current contour is the initial contour or the updated contour.

[0110] It can be understood that in the case of the embodied robot triggering a soft collision with the new obstacle, it means that the distance between the new obstacle and the current obstacle is relatively close, and at the same time it also means that the distance between the embodied robot and the new obstacle is relatively close. In order to avoid an actual collision with the new obstacle when avoiding the current obstacle, therefore, the embodied robot needs to identify the contour information of the new obstacle.

[0111] Exemplarily, in the case of the embodied robot triggering a soft collision with the new obstacle, the initial contour information of the new obstacle will be obtained, and then it is controlled to switch from the contour of the current obstacle to walking along the initial contour of the new obstacle. It should be understood that before switching the walking, multiple look-back points also need to be determined on the initial contour of the new obstacle according to the confirmation method of the look-back point as described above, and then when each corresponding look-back point is reached, the above-mentioned preset look-back action is executed to determine whether to perform the contour information update operation of the new obstacle until the process of walking along the entire contour of the new obstacle is completed.

[0112] It should be understood that the embodied robot in this embodiment can be a cleaning robot, a food delivery robot, a companion robot, a load-carrying robot, etc. If the embodied robot is a cleaning robot, during the process of avoiding obstacles along the edge by the method of this embodiment, not only can the garbage around the obstacle be cleaned up, but also during the process of bypassing the obstacle along the edge, the contour information of the obstacle can be recorded. When re-planning the path, the obstacle information will be considered, so as to avoid the obstacle. In addition, if the embodied robot is a food delivery robot, an autonomously driven load-carrying robot or a companion robot, etc., after avoiding obstacles and complementing the obstacle contour information by the obstacle avoidance and look-back method of this embodiment, when re-planning the path, the obstacle information will be considered, so as to avoid the obstacle.

[0113] In this embodiment, after the embodied robot recognizes an obstacle, it determines whether a soft collision occurs with the current obstacle. If a soft collision is triggered, a plurality of look-back points are determined on the initial polygon contour of the current obstacle recognized by the obstacle recognition module; then, the embodied robot is controlled to walk along the polygon contour. Each time a look-back point is reached, a preset look-back action is executed, and the contour information of the current obstacle is re-recognized; then, when the deviation between the re-recognized obstacle contour information and the previous contour information is large, the contour information of the current obstacle is updated, and new look-back points are determined on the updated contour and the process of walking along the contour is executed again until the walking along the contour of the current obstacle is completed or a new obstacle is encountered. Through this look-back operation, the actual contour of the obstacle can be re-acquired, the contour information of the obstacle can be continuously complemented during the walking process of the embodied robot, so that the embodied robot can better obtain the shape of the obstacle, and further make the embodied robot more stable and smooth when bypassing the obstacle next time, and can timely adjust the distance from the obstacle to avoid collision. In addition, since the contour of the obstacle is complemented, correspondingly, during subsequent path planning, the path can be optimized according to the complemented obstacle contour information to make the path more adaptable to the current environment, etc.

[0114] ​ FIG. shows a schematic structural diagram of an obstacle avoidance and look-back control device for an embodied robot according to an embodiment of the present application. Exemplarily, the obstacle avoidance and look-back control device for the embodied robot includes:

[0115] A determination module 100, configured to determine a plurality of look-back points on the initial contour according to the recognized initial contour information of the current obstacle and the position when the soft collision is triggered in the case of triggering a soft collision.

[0116] An execution module 200, configured to control the embodied robot to walk along the initial contour of the current obstacle, and execute a preset look-back action and obtain a corresponding look-back result each time a look-back point is reached.

[0117] An update module 300, configured to update the contour information of the current obstacle based on the look-back result that meets a preset condition.

[0118] In one implementation, the look-back result is the new contour information of the current obstacle re-identified after reaching a look-back point and performing a preset look-back action.

[0119] The update module 300 is specifically configured to update the contour information of the current obstacle according to the new contour information when the deviation between the new contour information identified at the current look-back point and the initial contour information exceeds a preset range.

[0120] In one implementation, the embodied robot obstacle avoidance look-back control device further includes:

[0121] A control module (not shown in the figure), configured to control the embodied robot to move to the next look-back point and perform a preset look-back action to determine whether to perform a contour information update operation when the deviation between the new contour information identified at the current look-back point and the initial contour information does not exceed a preset range.

[0122] In one implementation, after each execution of the contour information update operation, the determination module 100 is further configured to re-determine a plurality of new look-back points based on the updated contour information of the current obstacle; and the execution module 200 is further configured to control the embodied robot to walk along the updated contour, and perform a preset look-back action at each arrival at a new look-back point to determine whether to perform the contour information update operation again.

[0123] In one implementation, the control module is further configured to, during the process of controlling the embodied robot to walk along the current contour of the current obstacle, when not reaching the next look-back point, if a new obstacle is detected, and when no soft collision is triggered with the new obstacle, control the embodied robot to continue walking along the current contour of the current obstacle until reaching the next look-back point; where the current contour is the initial contour or the updated contour.

[0124] Further, the control module is further configured to, when a soft collision is triggered with the new obstacle, control the embodied robot to switch to walking along the initial contour of the new obstacle, and perform a preset look-back action at each arrival at a corresponding look-back point to determine whether to perform the contour information update operation of the new obstacle.

[0125] In one implementation, the determination module 100 is specifically configured to use the connection line between the position coordinate point at the time of triggering the soft collision and the center point of the initial contour of the current obstacle as a reference straight line, and select at least one target straight line that forms a preset angle with the reference straight line; then use the non-soft collision intersection point of the reference straight line and the initial contour, and all the intersection points of each target straight line and the initial contour as the selected plurality of look-back points.

[0126] In one embodiment, the embodied robot obstacle avoidance and look-back control device further includes:

[0127] An information collection module (not shown in the figure), configured to collect scene visual information through an obstacle avoidance sensor.

[0128] The determination module 100 is further configured to identify the obstacle type of the current obstacle according to the scene visual information and a preset obstacle type, and determine the initial contour information of the current obstacle based on the identified obstacle type; wherein, the initial contour information includes the polygon contour information of the current obstacle.

[0129] In one embodiment, the obstacle avoidance sensor is in the front of the embodied robot; performing a preset look-back action includes: controlling the embodied robot to rotate a specified angle and then retreat a preset distance, so that the current obstacle is within the viewing angle range of the obstacle avoidance sensor.

[0130] In one embodiment, the embodied robot obstacle avoidance and look-back control device further includes:

[0131] A soft collision detection module (not shown in the figure), configured to construct a prediction area in the movement direction of the embodied robot with the position of the embodied robot as the starting point; if the prediction area contains points on the initial contour of the current obstacle or points inside the initial contour, a soft collision is triggered.

[0132] It can be understood that the device in this embodiment corresponds to the embodied robot obstacle avoidance and look-back control method in the above embodiment, and the optional items in the above embodiment also apply to this embodiment, so they will not be described repeatedly here.

[0133] The present application also provides a computer-readable storage medium for storing the computer program used in the above-mentioned embodied robot. When the computer program is executed on a processor, the above-mentioned embodied robot obstacle avoidance and look-back control method is implemented. For example, the computer-readable storage medium may include, but is not limited to, various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0134] In several embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and structural diagrams in the accompanying drawings show the possible architectures, functions, and operations of devices, methods, and computer program products according to multiple embodiments of this application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in an alternative implementation, the functions marked in the blocks can occur in a different order from that marked in the accompanying drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the structural diagram and / or flowchart, as well as the combination of blocks in the structural diagram and / or flowchart, can be implemented by a dedicated hardware-based system that executes the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0135] In addition, each functional module or unit in various embodiments of this application can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0136] If the above functions are implemented in the form of software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application.

[0137] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered by the protection scope of this application.

Claims

1. An embodied robot obstacle avoidance and look-back control method, characterized in that, Including: In the case of triggering a soft collision, based on the initial contour information of the identified current obstacle and the position when the soft collision is triggered, determine a plurality of look-back points on the initial contour; wherein, the soft collision is a state where the distance between the embodied robot and the current obstacle is close but no real collision occurs; Control the embodied robot to walk along the initial contour of the current obstacle, and when reaching each look-back point, execute a preset look-back action and obtain a corresponding look-back result; wherein, the look-back result is the new contour information of the current obstacle re-identified after reaching a look-back point and executing the preset look-back action; In the case where the deviation between the new contour information identified at the current look-back point and the initial contour information exceeds a preset range, update the contour information of the current obstacle according to the new contour information; In the case where the deviation between the new contour information identified at the current look-back point and the initial contour information does not exceed the preset range, control the embodied robot to go to the next look-back point and execute the preset look-back action to determine whether to perform the contour information update operation; After each execution of the contour information update operation, based on the updated contour information of the current obstacle, re-determine a plurality of new look-back points; control the embodied robot to walk along the updated contour, and when reaching each new look-back point, execute the preset look-back action to determine whether to perform the contour information update operation again.

2. The method for obstacle avoidance and look-back control of an embodied robot according to claim 1, wherein, Further including: During the process of controlling the embodied robot to walk along the current contour of the current obstacle, when not reaching the next look-back point, if a new obstacle is detected, in the case of not triggering a soft collision with the new obstacle, control the embodied robot to continue walking along the current contour of the current obstacle until reaching the next look-back point; wherein, the current contour is the initial contour or the updated contour; In the case of triggering a soft collision with the new obstacle, control the embodied robot to switch to walking along the initial contour of the new obstacle, and when reaching each corresponding look-back point, execute the preset look-back action to determine whether to perform the contour information update operation of the new obstacle.

3. The embodied robot obstacle avoidance and look-back control method according to claim 1, wherein The step of determining a plurality of look-back points on the initial contour according to the initial contour information of the identified current obstacle and the position when the soft collision is triggered includes: Taking the line connecting the position coordinate point when the soft collision is triggered and the center point of the initial contour of the current obstacle as a reference line, and selecting at least one target line forming a preset angle with the reference line; Taking the non-soft collision intersection point of the reference line and the initial contour, and all the intersection points of each target line and the initial contour as the selected plurality of look-back points.

4. The embodied robot obstacle avoidance and look-back control method according to claim 1, characterized in that, Before the step of determining a plurality of look-back points on the initial contour according to the initial contour information of the identified current obstacle and the position when the soft collision is triggered, further including: Collect scene visual information through an obstacle avoidance sensor. Identify the obstacle type of the current obstacle according to the scene visual information and the preset obstacle types, and determine the initial contour information of the current obstacle based on the identified obstacle type; wherein, the initial contour information includes the polygon contour information of the current obstacle.

5. The embodied robot obstacle avoidance and look-back control method according to claim 4, wherein The obstacle avoidance sensor is in the direct front of the embodied robot; the execution of the preset look-back action includes: Controlling the embodied robot to rotate by a specified angle and then retreat by a preset distance, so that the current obstacle is within the viewing angle range of the obstacle avoidance sensor.

6. The method for controlling obstacle avoidance and look-back of an embodied robot according to claim 1, characterized in that, The determination process of triggering the soft collision includes: Taking the position of the embodied robot as the starting point, construct a prediction area in the moving direction of the embodied robot; If the prediction area contains points on the initial contour of the current obstacle or points inside the initial contour, trigger a soft collision.

7. The method for controlling obstacle avoidance and look-back of an embodied robot according to claim 6, wherein Taking the position of the embodied robot as the starting point, constructing a prediction area in the moving direction of the embodied robot includes: Taking the center point coordinates of the embodied robot as the starting point, construct a sector area with a radius of a predetermined length and a central angle of a predetermined angle in the moving direction of the embodied robot, and use the sector area as the prediction area.

8. The embodied robot obstacle avoidance and look-back control method according to claim 3, wherein, The preset angles include one or more combinations of 90 degrees, 45 degrees, 60 degrees, and 135 degrees.

9. An embodied robot obstacle avoidance and look-back control device, characterized in that, Includes: A determination module, configured to, in the case of triggering a soft collision, determine a plurality of look-back points on the initial contour according to the initial contour information of the identified current obstacle and the position when the soft collision is triggered; wherein, the soft collision is a state where the distance between the embodied robot and the current obstacle is close but no real collision occurs; An execution module, configured to control the embodied robot to walk along the initial contour of the current obstacle, and when reaching each of the look-back points, execute the preset look-back action and obtain the corresponding look-back result; wherein, the look-back result is the new contour information of the current obstacle re-identified after reaching a look-back point and executing the preset look-back action; An update module, configured to update the contour information of the current obstacle according to the new contour information when the deviation between the new contour information identified at the current look-back point and the initial contour information exceeds a preset range; A control module, configured to, when the deviation between the new contour information identified at the current look-back point and the initial contour information does not exceed the preset range, control the embodied robot to go to the next look-back point and execute the preset look-back action to determine whether to perform the contour information update operation; After each execution of the contour information update operation, the determination module is further configured to re-determine a plurality of new look-back points based on the updated contour information of the current obstacle; and, the execution module is further configured to control the embodied robot to walk along the updated contour, and when reaching each of the new look-back points, execute the preset look-back action to determine whether to perform the contour information update operation again.

10. An embodied robot system, characterized in that, Includes: An actuator for performing corresponding actions; A memory, where the memory stores a computer program; A processor, the processor being configured to execute the computer program to implement the embodied robot obstacle avoidance and look-back control method according to any one of claims 1-8.

Images