Quadruped robot path planning method and device based on large model and storage medium

Abstract

The invention discloses a quadruped robot path planning method and device based on a large model and a storage medium, and relates to the technical field of quadruped robots. Obtaining a preferred treading point area, a warning treading point area and a dangerous treading point area based on the motion contrast data; the to-be-planned path is adjusted, and the robot is controlled to move according to the standard motion path; the method is used for solving the problems that an existing robot path planning method lacks a targeted path planning method when the quadruped robot moves to various terrains, so that after path planning, the quadruped robot has abnormal behaviors which influence normal movement, such as unstable gravity center or toppling and the like, at unknown positions of different terrains; and therefore, the quadruped robot cannot effectively complete the moving target.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of quadruped robots, in particular to a quadruped robot path planning method based on a large model, equipment and a storage medium. BACKGROUND

[0002] A quadruped robot is a kind of bionic robot that moves through four legs, simulates the movement of quadruped animals, has complex terrain adaptability, and is widely used in military, rescue and industrial fields; quadruped robot path planning refers to determining the optimal or feasible motion trajectory of the robot from the starting point to the target point in a complex environment through an algorithm, and the core is to balance the motion efficiency, stability and terrain adaptability.

[0003] The existing method for quadruped robot path planning usually places the path to be planned and the heading angle in a three-dimensional map to realize the conversion of a two-dimensional navigation task into a three-dimensional space search task, and then simplifies the geometric shape of the quadruped robot to improve the path planning efficiency and obtain a smoother path. Although this improvement method can ensure that the path planning result is not too conservative and improve the efficiency of obtaining a smooth path, it lacks a targeted path planning method when the quadruped robot moves to various terrains, resulting in abnormal behaviors such as unstable center of gravity or tipping of the quadruped robot at unknown positions in different terrains after path planning, which affects normal motion, thereby making the quadruped robot unable to effectively complete the motion target. For example, in the patent application with publication number CN115167425A, a quadruped robot map construction and path planning method is disclosed, which is to represent the position to be planned and the heading angle in a three-dimensional grid map to convert a two-dimensional plane navigation task into a three-dimensional space search problem, improve the efficiency of path planning, and simplify the geometric shape of the quadruped robot to construct unique kinematic constraints and plan a smooth path. Other methods for quadruped robot path planning usually improve obstacle avoidance, but still cannot solve the problem of lacking a targeted path planning method when the quadruped robot moves to various terrains, resulting in abnormal behaviors such as unstable center of gravity or tipping of the quadruped robot at unknown positions in different terrains after path planning, which affects normal motion, thereby making the quadruped robot unable to effectively complete the motion target. Therefore, it is necessary to improve the existing quadruped robot path planning method. SUMMARY

[0004] The application aims to at least solve one of the technical problems in the prior art by proposing a large model-based quadruped robot path planning method, device and storage medium, to solve the problem that the existing robot path planning method lacks a targeted path planning method when the quadruped robot moves to various terrains, resulting in abnormal behaviors such as unstable center of gravity or tipping of the quadruped robot at unknown positions of different terrains after path planning, affecting normal movement, and thus the quadruped robot cannot effectively complete the movement target.

[0005] To achieve the above-mentioned purpose, in a first aspect, the application provides a large model-based quadruped robot path planning method, comprising the following steps: Obtaining motion data of the quadruped robot based on a large model, denoted as motion analysis data; screening the motion analysis data, and obtaining a plurality of quadruped motion terrains and motion reference data of each quadruped motion terrain based on the screening result, wherein the motion reference data includes regular data and accident data; Analyzing the motion reference data of each quadruped motion terrain, and obtaining the preferred foothold area, warning foothold area and dangerous foothold area corresponding to each quadruped motion terrain based on the analysis result; Obtaining a path to be planned, segmenting the path to be planned based on all quadruped motion terrains, and obtaining a plurality of segmented planning paths; adjusting each segmented planning path based on the preferred foothold area, warning foothold area and dangerous foothold area of the quadruped motion terrain, and denoting the path obtained by restoring the adjusted segmented planning path as a standard motion path of the quadruped robot, to control the robot motion with the standard motion path.

[0006] Further, obtaining motion data of the quadruped robot based on a large model, denoted as motion analysis data; screening the motion analysis data, and obtaining a plurality of quadruped motion terrains and motion reference data of each quadruped motion terrain based on the screening result, including: Obtaining data corresponding to the motion analysis data in the sensor data of the quadruped robot, denoted as motion sensor data; analyzing the motion sensor data using a large model, and respectively obtaining data corresponding to uphill, downhill, straight motion and turning of the quadruped robot in the motion sensor data, denoted as uphill sensor data, downhill sensor data, straight line sensor data and turning sensor data, wherein the quadruped motion terrain includes uphill terrain, downhill terrain, straight motion terrain and turning terrain.

[0007] Further, screening the motion analysis data, and obtaining a plurality of quadruped motion terrains and motion reference data of each quadruped motion terrain based on the screening result further includes: For uphill sensing data: data corresponding to the uphill sensing data in the motion analysis data is recorded as uphill analysis data of the uphill terrain, wherein the uphill analysis data contains multiple sets of data, and each set of data corresponds to an uphill motion of the quadruped robot; Data in the multiple sets of data of the uphill analysis data that exist abnormal behavior warning is recorded as accident data, and data in the multiple sets of data of the uphill analysis data that do not exist abnormal behavior warning is recorded as normal data.

[0008] Further, the motion contrast data of each quadruped motion terrain is analyzed, and based on the analysis result, the preferred foothold area, the warning foothold area and the dangerous foothold area corresponding to each quadruped motion terrain are obtained, including: For any one set of accident data: based on the uphill sensing data corresponding to the accident data, the plan view of the path under the overhead angle in the uphill motion corresponding to the accident data of the quadruped robot is obtained and recorded as the overhead motion plan view; the minimum circumscribed rectangle of the overhead motion plan view is obtained and recorded as the landing point analysis rectangle; all landing points of the four mechanical feet of the quadruped robot in the uphill motion corresponding to the accident data are obtained and marked as mechanical landing points in the landing point analysis rectangle; The mechanical landing point generated when the abnormal behavior data in the accident data is uploaded is obtained from all mechanical landing points and recorded as the accident landing point; the mechanical landing point other than the accident landing point is recorded as the normal landing point; the maximum contour fitted by all normal landing points is obtained, and the area within the contour is recorded as the normal landing point area; the maximum contour fitted by all accident landing points is obtained, and the area within the contour other than the normal landing point area is recorded as the accident landing point area.

[0009] Further, the motion contrast data of each quadruped motion terrain is analyzed, and based on the analysis result, the preferred foothold area, the warning foothold area and the dangerous foothold area corresponding to each quadruped motion terrain are obtained, including: For any one set of normal data: based on the analysis method of the accident data, the landing point analysis rectangle corresponding to the normal data and all mechanical landing points in the landing point analysis rectangle are obtained, and the area within the maximum contour fitted by all mechanical landing points is recorded as the normal landing point area; The normal landing point area and the accident landing point area of all accident data, and the normal landing point area of all normal data are obtained.

[0010] Further, the motion contrast data of each quadruped motion terrain is analyzed, and based on the analysis result, the preferred foothold area, the warning foothold area and the dangerous foothold area corresponding to each quadruped motion terrain are obtained, including: randomly obtaining all accident data and a set of data α from the regular data; taking the landing point analysis rectangle of data α as a standard rectangle; for all accident data except data α: scaling the landing point analysis rectangle of accident data, and the regular landing point area and accident landing point area in the landing point analysis rectangle proportionally based on the standard rectangle, until the landing point analysis rectangle of accident data is the same as the standard rectangle, wherein the motion direction of the quadruped robot in the landing point analysis rectangle is kept consistent during the scaling; for all regular data except data α: scaling the landing point analysis rectangle of regular data proportionally based on the standard rectangle, until the landing point analysis rectangle of regular data is the same as the standard rectangle; completely overlapping all scaled landing point analysis rectangles with the standard rectangle, and taking the obtained standard rectangle as a landing point overlap rectangle, wherein the motion direction of the quadruped robot in all landing point analysis rectangles is kept consistent during the overlapping.

[0011] Further, analyzing the motion contrast data of each quadruped motion terrain, and obtaining the preferred foot point area, warning foot point area and dangerous foot point area corresponding to each quadruped motion terrain based on the analysis results further comprises: taking the area in the landing point overlap rectangle where only the regular landing point area exists as the preferred foot point area of uphill motion, taking the area in the landing point overlap rectangle where only the accident landing point area exists as the dangerous foot point area of uphill motion; taking the area in the landing point overlap rectangle where the regular landing point area and the accident landing point area overlap as the warning foot point area of uphill motion; Based on the analysis method of uphill sensing data, the downhill sensing data, the straight line sensing data and the turning sensing data are analyzed respectively, and the preferred foot point area, the warning foot point area and the dangerous foot point area corresponding to the downhill terrain, the straight line motion terrain and the turning terrain are obtained.

[0012] Further, obtaining a to-be-planned path, segmenting the to-be-planned path based on all quadruped motion terrains, and obtaining a plurality of segmented planning paths; adjusting each segmented planning path based on the preferred foot point area, the warning foot point area and the dangerous foot point area of the quadruped motion terrain, and taking the path obtained by restoring the adjusted segmented planning path as the standard motion path of the quadruped robot, and controlling the robot motion based on the standard motion path comprises: Before a quadruped robot performs a movement, a path of the quadruped robot performing the movement is obtained, denoted as a to-be-planned path; a large model is used to perform segmentation processing on the to-be-planned path, and the segmentation processing includes: using the large model to simulate movement of the quadruped robot in the to-be-planned path, and recording obtained sensor data as simulated sensor data; based on an analysis manner of the large model when analyzing the movement sensor data, using the large model to obtain uphill terrain, downhill terrain, straight motion terrain and turning terrain in the to-be-planned path from the simulated sensor data, and respectively marking as segmented planning paths; For any one segmented planning path: based on the simulated sensor data, a plan view of a path of the simulated quadruped robot in the segmented planning path is obtained under a top-down angle, and is denoted as a simulated top-down plan view; a minimum circumscribed rectangle of the simulated top-down plan view is denoted as a simulated top-down rectangle; When the segmented planning path is the uphill terrain, the simulated top-down rectangle is scaled in proportion based on the landing point coincident rectangle corresponding to the uphill movement as a reference, until the landing point coincident rectangle and the simulated top-down rectangle can completely coincide when the motion direction of the quadruped robot in the simulated top-down rectangle is the same as the landing point coincident rectangle; the same preferred foothold area, dangerous foothold area and warning foothold area as the landing point coincident rectangle are marked in the simulated top-down rectangle after scaling in proportion; The simulated top-down rectangle is restored to a state before scaling in proportion, and the preferred foothold area, the dangerous foothold area and the warning foothold area in the simulated top-down rectangle are scaled in the same proportion based on the proportion of scaling in proportion of the simulated top-down rectangle when the simulated top-down rectangle is restored; Based on the analysis manner when the segmented planning path is the uphill terrain, the same manner of analysis is performed when the segmented planning path is the downhill terrain, the straight motion terrain or the turning terrain, and the corresponding simulated top-down rectangle and the preferred foothold area, the dangerous foothold area and the warning foothold area in the simulated top-down rectangle are obtained; Based on a position of each segmented planning path in the to-be-planned path, a path formed by the preferred foothold areas of all the segmented planning paths is denoted as a standard motion path of the quadruped robot, and the quadruped robot performs movement along the standard motion path when performing movement; when a landing point of any one mechanical foot of the quadruped robot during actual movement is in the warning foothold area of any one segmented planning path, a path deviation warning is sent until the landing points of all the mechanical feet of the quadruped robot are in the preferred foothold area; When the landing point of any one mechanical foot during actual movement is in the dangerous foothold area of any one segmented planning path, the movement of the quadruped robot is stopped and is reported to a worker.

[0013] In a second aspect, the present application provides an electronic device comprising a processor and a memory, wherein the memory stores computer readable instructions which, when executed by the processor, perform the steps of the above method.

[0014] In a third aspect, the present application provides a storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the above method.

[0015] The present application has the following advantages: firstly, the present application obtains motion analysis data based on a large model; the motion analysis data is filtered, and based on the filtering result, a plurality of quadruped motion terrains and motion contrast data of each quadruped motion terrain in the motion analysis data are obtained; then, the motion contrast data of each quadruped motion terrain is analyzed, and based on the analysis result, a preferred foothold area, a warning foothold area and a dangerous foothold area corresponding to each quadruped motion terrain are obtained, which has the advantage that by obtaining the motion contrast data of a plurality of quadruped motion terrains and obtaining the corresponding preferred foothold area, warning foothold area and dangerous foothold area, based on the existing motion data of the quadruped robot, the foothold of the mechanical foot of the quadruped robot when the quadruped robot moves to each terrain, can normally move, has a low probability of abnormal behavior and a high probability of abnormal behavior, so as to facilitate effective adjustment of the path of different terrains in the subsequent path planning based on the preferred foothold area, the warning foothold area and the dangerous foothold area, thereby reducing the probability of abnormal behavior of the quadruped robot when moving along the planned path to each terrain, and further ensuring that the quadruped robot can effectively complete the motion target. The present application also obtains a to-be-planned path, and performs segmentation processing on the to-be-planned path based on all the quadruped motion terrains; each segmented planning path is adjusted based on the preferred foothold area, the warning foothold area and the dangerous foothold area of the quadruped motion terrain, and the path obtained by restoring the adjusted segmented planning path is recorded as a standard motion path of the quadruped robot, and the robot is controlled to move along the standard motion path, which has the advantage that by performing segmentation processing on the to-be-planned path, different terrains in the to-be-planned path can be separated, so as to adjust the path corresponding to each terrain in the to-be-planned path, so as to ensure that the quadruped robot can effectively complete the motion target after the adjusted path is restored. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of the steps of the method of the present application; Figure 2 An acquisition diagram of the landing point analysis rectangle of the present application; Figure 3 An acquisition diagram of the conventional landing point area and the accident landing point area of the present application; Figure 4A schematic diagram of a path to be planned of the present application; Figure 5 A schematic diagram of a segmented planning path corresponding to the path to be planned of the present application; Figure 6 A structural schematic diagram of the electronic device of the present application. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the 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. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0018] Embodiment 1, please refer to Figure 1 As shown in the figure, the present application provides a quadruped robot path planning method based on a large model, including the following steps: Step S1, obtaining motion data of a quadruped robot based on a large model, denoted as motion analysis data; screening the motion analysis data, and obtaining a plurality of quadruped motion terrains and motion comparison data of each quadruped motion terrain based on the screening result, wherein the motion comparison data includes regular data and accident data; Step S1 includes: step S101, obtaining data corresponding to the motion analysis data in the sensor data of the quadruped robot, and denoted as motion sensor data; using a large model to analyze the motion sensor data, and respectively obtaining data corresponding to uphill, downhill, straight motion and turning of the quadruped robot in the motion sensor data, respectively denoted as uphill sensor data, downhill sensor data, straight line sensor data and turning sensor data, wherein the quadruped motion terrain includes uphill terrain, downhill terrain, straight motion terrain and turning terrain; In the specific implementation process, if there are many different terrains other than uphill terrain, downhill terrain, straight motion terrain and turning terrain in actual analysis, such as stair-shaped uphill and slope-shaped uphill, the quadruped motion terrain is divided more carefully, so as to ensure that the quadruped motion terrain can contain all terrains in the path to be planned, so as to realize more comprehensive path planning; Step S102, for the uphill sensor data: the data corresponding to the uphill sensor data in the motion analysis data is denoted as uphill analysis data of the uphill terrain, wherein the uphill analysis data contains a plurality of groups of data, and each group of data corresponds to one uphill motion of the quadruped robot; In the implementation process, for example, in the uphill sensing data, the height of the quadruped robot starts to rise compared to the horizontal plane at time t1, and continues until the height of the quadruped robot stops rising compared to the horizontal plane at time t2, then the data between time t1 and t2 in the data corresponding to the motion analysis data and the uphill sensing data of the uphill terrain can be recorded as the uphill sensing data of the uphill terrain. Step S103, record the data with abnormal behavior warning in the multiple sets of data of the uphill analysis data as accident data, and record the data without abnormal behavior warning in the multiple sets of data of the uphill analysis data as regular data.

[0019] Step S2, analyze the motion contrast data of each quadruped motion terrain, and obtain the preferred foot point area, warning foot point area and dangerous foot point area corresponding to each quadruped motion terrain based on the analysis result; Step S2 includes: step S201, for any one set of accident data: based on the uphill sensing data corresponding to the accident data, obtain the plan view of the path under the overhead angle in the uphill motion corresponding to the accident data, and record it as the overhead motion plan view; obtain the minimum circumscribed rectangle of the overhead motion plan view, and record it as the landing point analysis rectangle; obtain all landing points of the four mechanical feet of the quadruped robot in the uphill motion corresponding to the accident data, and mark them as mechanical landing points in the landing point analysis rectangle; In the implementation process, for example, in one data analysis, the path of the uphill motion of the quadruped robot is as shown by the curve CC1 between the point AA1 and the point BB1 in Figure 2 , then the overhead motion plan view obtained by obtaining the plan view of the path under the overhead angle is DD1, and since DD1 is a rectangle, the corresponding landing point analysis rectangle is also DD1; by obtaining the landing point analysis rectangle, it can be ensured that when there are multiple different uphill paths, the landing point analysis rectangle is obtained uniformly to analyze all uphill paths equally, so as to obtain the preferred foot point area, warning foot point area and dangerous foot point area corresponding to all uphill paths, and ensure more comprehensive data analysis.

[0020] Step S202, obtain the mechanical landing point generated when the abnormal behavior data in the accident data is uploaded among all mechanical landing points, and record it as the accident landing point; record the mechanical landing points other than the accident landing points as the regular landing points; obtain the maximum contour fitted by all regular landing points, and record the area within the contour as the regular landing point area; obtain the maximum contour fitted by all accident landing points, and record the area within the contour other than the regular landing point area as the accident landing point area; In the implementation process, for example, in one data analysis, all landing points in the landing point analysis rectangle DD1 obtained are as shown by the curve CC1 between the point AA1 and the point BB1 in Figure 3As shown, the center of all Δs is the accident drop point, and the center of all Os is the normal drop point. Through analysis, it is obtained that the region EE1 is the normal drop point region, and the regions FF1 and FF2 are the accident drop point regions.

[0021] Step S2 further includes: step S203, for any one group of normal data: based on the analysis manner of the accident data, obtaining the drop point analysis rectangle corresponding to the normal data and all the mechanical drop points in the drop point analysis rectangle, and recording the region in the maximum contour obtained by fitting all the mechanical drop points as the normal drop point region; Step S204, obtaining the normal drop point region and the accident drop point region of all the accident data, and the normal drop point region of all the normal data.

[0022] Step S2 further includes: step S205, randomly obtaining a group of data a from all the accident data and the normal data; recording the drop point analysis rectangle of the data a as the standard rectangle; for all the accident data except the data a: scaling the drop point analysis rectangle of the accident data, and the normal drop point region and the accident drop point region in the drop point analysis rectangle in the same proportion based on the standard rectangle, until the drop point analysis rectangle of the accident data is the same as the standard rectangle, wherein the same proportion scaling maintains the same motion direction of the quadruped robot in the standard rectangle and the drop point analysis rectangle; Step S206, for all the normal data except the data a: scaling the drop point analysis rectangle of the normal data in the same proportion based on the standard rectangle, until the drop point analysis rectangle of the normal data is the same as the standard rectangle; Step S207, completely coinciding all the scaled drop point analysis rectangles with the standard rectangle, and recording the obtained standard rectangle as the drop point coincidence rectangle, wherein the coincidence maintains the same motion direction of the quadruped robot in all the drop point analysis rectangles; In the specific implementation process, by obtaining the drop point coincidence rectangle, the normal drop point region and the accident drop point region corresponding to all the paths of the quadruped robot performing uphill motion can be integrated, so as to ensure that the obtained preferred foothold region, warning foothold region and dangerous foothold region are the results obtained by combining all the uphill motions of the quadruped robot, and can better adapt to the motion habits of the quadruped robot itself.

[0023] Step S2 further includes: step S208, recording the region in the drop point coincidence rectangle where only the normal drop point region exists as the preferred foothold region of the uphill motion, recording the region in the drop point coincidence rectangle where only the accident drop point region exists as the dangerous foothold region of the uphill motion, and recording the region in the drop point coincidence rectangle where the normal drop point region and the accident drop point region coincide as the warning foothold region of the uphill motion; Step S209, based on the analysis mode of the uphill sensing data, the downhill sensing data, the straight sensing data and the turning sensing data are analyzed respectively, and the preferred foothold area, the warning foothold area and the dangerous foothold area corresponding to the downhill terrain, the straight motion terrain and the turning terrain are obtained.

[0024] Step S3, obtain the to-be-planned path, segment the to-be-planned path based on all the quadruped motion terrains, and obtain a plurality of segmented planning paths; adjust each segmented planning path based on the preferred foothold area, the warning foothold area and the dangerous foothold area of the quadruped motion terrain, and record the path restored from the adjusted segmented planning path as the standard motion path of the quadruped robot, and control the robot motion with the standard motion path; Step S3 includes: step S301, before the quadruped robot performs motion, obtaining the path of the quadruped robot performing motion, recorded as the to-be-planned path; using a large model to segment the to-be-planned path, the segmentation processing includes: using the large model to simulate the motion of the quadruped robot in the to-be-planned path, and recording the obtained sensor data as simulation sensor data; based on the analysis mode when the large model analyzes the motion sensor data, using the large model to obtain the uphill terrain, the downhill terrain, the straight motion terrain and the turning terrain in the to-be-planned path from the simulation sensor data, and mark them as segmented planning paths respectively; In the specific implementation process, for example, in one data analysis, the to-be-planned path of the quadruped robot obtained is as shown by the curve CC2 between AA2 and BB2 in Figure 4 , then by segmenting the uphill terrain, the downhill terrain, the straight motion terrain and the turning terrain in the to-be-planned path respectively, the paths corresponding to the uphill terrain, the downhill terrain, the straight motion terrain and the turning terrain are respectively as shown by the paths in TT1 to TT4 in Figure 5 , then at this time, the paths in TT1 to TT4 can be recorded as segmented planning paths and subsequent analysis is performed; Step S302, for any one segmented planning path: based on the simulation sensor data, obtain the plan view of the path under the overhead angle when the simulation quadruped robot moves in the segmented planning path, and record it as the simulation overhead plan view; record the minimum circumscribed rectangle of the simulation overhead plan view as the simulation overhead rectangle; Step S303, when the segmented planning path is the uphill terrain, take the landing point coincidence rectangle corresponding to the uphill motion as the reference, proportionally scale the simulation overhead rectangle until the landing point coincidence rectangle and the overhead simulation rectangle are completely coincident when the motion direction of the quadruped robot in the overhead simulation rectangle is the same as the landing point coincidence rectangle; mark the preferred foothold area, the dangerous foothold area and the warning foothold area which are completely the same as the landing point coincidence rectangle in the simulation overhead rectangle after the proportional scaling; Step S304, restore the simulated overhead rectangle to the state before the equal proportion scaling, and based on the proportion of the equal proportion scaling of the simulated overhead rectangle when being restored, scale the preferred foot point area, the dangerous foot point area and the warning foot point area in the simulated overhead rectangle by the same proportion; In the specific implementation process, by placing the preferred foot point area, the dangerous foot point area and the warning foot point area of each path in all the segmented planning paths in the same proportion, the standard motion path obtained can be ensured to be the motion trajectory with the minimum probability of abnormal behavior of the quadruped robot when moving in the planning path, and the real-time motion of the quadruped robot can also be monitored through the dangerous foot point area and the warning foot point area to timely give a danger warning when the quadruped robot moves to the area where the abnormal behavior is prone to occur; Step S305, based on the analysis mode when the segmented planning path is an uphill terrain, analyze in the same way when the segmented planning path is a downhill terrain, a straight motion terrain or a turning terrain, and obtain the corresponding simulated overhead rectangle and the preferred foot point area, the dangerous foot point area and the warning foot point area in the simulated overhead rectangle; Step S306, based on the position of each segmented planning path in the planning path, record the path composed of the preferred foot point areas of all the segmented planning paths as the standard motion path of the quadruped robot, and execute the motion along the standard motion path when the quadruped robot executes the motion; when the landing point of any one of the mechanical feet of the quadruped robot is in the warning foot point area of any one of the segmented planning paths when actually moving, send a path deviation warning until the landing points of all the mechanical feet of the quadruped robot are in the preferred foot point area; Step S307, when the landing point of any one of the mechanical feet when actually moving is in the dangerous foot point area of any one of the segmented planning paths, stop the motion of the quadruped robot and report to the staff.

[0025] Embodiment 2, please refer to Figure 6 as shown, Figure 6An example is provided for a structural diagram of an electronic device, which can include a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete the communication among each other through the communication bus. The memory stores computer readable instructions, and the processor can invoke the instructions in the memory. When the computer readable instructions are executed by the processor, the steps in the method for path planning of a quadruped robot based on a large model are run to realize the following functions: first, obtaining motion analysis data based on a large model; screening the motion analysis data, and obtaining a plurality of quadruped motion terrains and motion reference data of each quadruped motion terrain based on the screening result; then analyzing the motion reference data of each quadruped motion terrain, and obtaining a preferred foot point area, a warning foot point area and a dangerous foot point area corresponding to each quadruped motion terrain based on the analysis result; finally, obtaining a to-be-planned path, performing segmentation processing on the to-be-planned path based on all the quadruped motion terrains, and obtaining a plurality of segmented planning paths; adjusting each segmented planning path based on the preferred foot point area, the warning foot point area and the dangerous foot point area of the quadruped motion terrain, and recording the path obtained by restoring the adjusted segmented planning path as a standard motion path of the quadruped robot, to control the robot motion with the standard motion path.

[0026] In addition, the logical instructions in the memory described above can be implemented in the form of a software function unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes: a U disk, 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 disk, and various program code storage media.

[0027] In embodiment 3, the present application also provides a computer program product, which comprises a computer program stored on a computer readable storage medium, and the computer program comprises program instructions, when the program instructions are executed by a computer, the computer can execute the four-legged robot path planning method based on a large model provided by each method, and the method comprises the following steps: first, acquiring motion analysis data based on a large model; screening the motion analysis data, and acquiring a plurality of four-legged motion terrains and motion reference data of each four-legged motion terrain based on the screening result; then, analyzing the motion reference data of each four-legged motion terrain, and acquiring the preferred foothold area, the warning foothold area and the dangerous foothold area corresponding to each four-legged motion terrain based on the analysis result; finally, acquiring a to-be-planned path, performing segmentation processing on the to-be-planned path based on all four-legged motion terrains, and obtaining a plurality of segmented planning paths; adjusting each segmented planning path based on the preferred foothold area, the warning foothold area and the dangerous foothold area of the four-legged motion terrain, and recording the path restored from the adjusted segmented planning path as the standard motion path of the four-legged robot, so as to control the robot motion by using the standard motion path.

[0028] In embodiment 4, the present application also provides a computer readable storage medium, and the present application provides a storage medium, which stores a computer program, and the computer program is executed by a processor to run the steps in the four-legged robot path planning method based on a large model, so as to realize the following functions: first, acquiring motion analysis data based on a large model; screening the motion analysis data, and acquiring a plurality of four-legged motion terrains and motion reference data of each four-legged motion terrain based on the screening result; then, analyzing the motion reference data of each four-legged motion terrain, and acquiring the preferred foothold area, the warning foothold area and the dangerous foothold area corresponding to each four-legged motion terrain based on the analysis result; finally, acquiring a to-be-planned path, performing segmentation processing on the to-be-planned path based on all four-legged motion terrains, and obtaining a plurality of segmented planning paths; adjusting each segmented planning path based on the preferred foothold area, the warning foothold area and the dangerous foothold area of the four-legged motion terrain, and recording the path restored from the adjusted segmented planning path as the standard motion path of the four-legged robot, so as to control the robot motion by using the standard motion path.

[0029] Through the description of the above embodiments, the embodiments of the present application can be provided as a method, a system or a computer program product. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the method described in each embodiment or some parts of the embodiment.

[0030] In the embodiments provided by the present application, it should be understood that the disclosed system or method can be implemented in other manners. The embodiments described above are merely specific implementation manners of the present application, and are not intended to limit the protection scope of the present application. For example, the division of the modules or the units is only a logical function division, and there can be another division manner in actual implementation. For example, a plurality of modules or units can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different parts can be indirect couplings or communication connections through some interfaces, communication interfaces, or buses, and can be in electrical, mechanical, or other forms.

[0031] Finally, it should be noted that the above embodiments are merely used to illustrate the technical solutions of the present application, rather than limit them; even though the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still make modifications to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A quadruped robot path planning method based on a large model, characterized in that, The method comprises the following steps: Obtain motion data of the quadruped robot based on a large model, denoted as motion analysis data; filter the motion analysis data, and obtain a plurality of quadruped motion terrains and motion contrast data of each quadruped motion terrain based on the filtering result, wherein the motion contrast data comprises regular data and accident data; Analyze the motion contrast data of each quadruped motion terrain, and obtain the preferred foothold area, warning foothold area and dangerous foothold area corresponding to each quadruped motion terrain based on the analysis result; Obtain a to-be-planned path, segment the to-be-planned path based on all quadruped motion terrains, and obtain a plurality of segmented planning paths; adjust each segmented planning path based on the preferred foothold area, warning foothold area and dangerous foothold area of the quadruped motion terrain, and denote the path restored from the adjusted segmented planning path as a standard motion path of the quadruped robot, so as to control the robot motion by using the standard motion path.

2. The large model-based quadruped robot path planning method according to claim 1, wherein, Obtain motion data of the quadruped robot based on a large model, denoted as motion analysis data; filter the motion analysis data, and obtain a plurality of quadruped motion terrains and motion contrast data of each quadruped motion terrain based on the filtering result, wherein the motion contrast data comprises regular data and accident data; Obtain data corresponding to the motion analysis data in the sensor data of the quadruped robot, denoted as motion sensor data; analyze the motion sensor data by using a large model, and obtain data corresponding to uphill, downhill, straight motion and turning of the quadruped robot in the motion sensor data, respectively denoted as uphill sensor data, downhill sensor data, straight line sensor data and turning sensor data, wherein the quadruped motion terrain comprises uphill terrain, downhill terrain, straight motion terrain and turning terrain.

3. The large model-based quadruped robot path planning method according to claim 2, characterized in that, Filter the motion analysis data, and obtain a plurality of quadruped motion terrains and motion contrast data of each quadruped motion terrain based on the filtering result, wherein the method further comprises: For the uphill sensor data: data corresponding to the uphill sensor data in the motion analysis data is denoted as uphill analysis data of the uphill terrain, wherein the uphill analysis data comprises a plurality of groups of data, and each group of data corresponds to one uphill motion of the quadruped robot; Data in the plurality of groups of data of the uphill analysis data that exist abnormal behavior warning are denoted as accident data, and data in the plurality of groups of data of the uphill analysis data that do not exist abnormal behavior warning are denoted as regular data.

4. The large model-based quadruped robot path planning method according to claim 3, characterized in that, Analyze the motion contrast data of each quadruped motion terrain, and obtain the preferred foothold area, warning foothold area and dangerous foothold area corresponding to each quadruped motion terrain based on the analysis result, wherein the method further comprises: For any one group of accident data: based on the uphill sensor data corresponding to the accident data, a plan view of a path of the quadruped robot in the uphill motion corresponding to the accident data is obtained from a top view, and denoted as a top view motion plan view; a minimum circumscribed rectangle of the top view motion plan view is obtained, denoted as a landing point analysis rectangle; all landing points of the four mechanical feet of the quadruped robot in the uphill motion corresponding to the accident data are obtained, and marked as robot landing points in the landing point analysis rectangle; Obtain all mechanical landing points generated when abnormal behavior data in the accident data is uploaded, and mark as accident landing points; mark the mechanical landing points other than the accident landing points as regular landing points; obtain the largest contour fitted by all regular landing points, and mark the area within the contour as a regular landing point area; obtain the largest contour fitted by all accident landing points, and mark the area within the contour other than the regular landing point area as an accident landing point area.

5. The large model-based quadruped robot path planning method according to claim 4, characterized in that, The analysis of the motion contrast data of each quadruped motion terrain and the obtaining of the preferred foot point area, the warning foot point area and the dangerous foot point area corresponding to each quadruped motion terrain based on the analysis result further comprise: For any one group of regular data: based on the analysis method of the accident data, obtain the landing point analysis rectangle corresponding to the regular data and all mechanical landing points within the landing point analysis rectangle, and mark the area within the largest contour fitted by all mechanical landing points as a regular landing point area; Obtain the regular landing point area and the accident landing point area of all accident data, and the regular landing point area of all regular data.

6. The large model-based quadruped robot path planning method according to claim 5, characterized in that, The analysis of the motion contrast data of each quadruped motion terrain and the obtaining of the preferred foot point area, the warning foot point area and the dangerous foot point area corresponding to each quadruped motion terrain based on the analysis result further comprise: Randomly obtain a group of data α from all accident data and regular data; mark the landing point analysis rectangle of data α as a standard rectangle; for all accident data other than data α: scale the landing point analysis rectangle of the accident data, and the regular landing point area and the accident landing point area within the landing point analysis rectangle in the same proportion based on the standard rectangle, until the landing point analysis rectangle of the accident data is the same as the standard rectangle, wherein the motion direction of the quadruped robot within the standard rectangle and the landing point analysis rectangle is kept consistent during the scaling; For all regular data other than data α: scale the landing point analysis rectangle of the regular data in the same proportion based on the standard rectangle, until the landing point analysis rectangle of the regular data is the same as the standard rectangle; Make all the scaled landing point analysis rectangles coincide with the standard rectangle completely, and mark the obtained standard rectangle as a landing point coincidence rectangle, wherein the motion direction of the quadruped robot within all the landing point analysis rectangles is kept consistent during the coincidence.

7. The large model-based quadruped robot path planning method according to claim 6, characterized in that, The analysis of the motion contrast data of each quadruped motion terrain and the obtaining of the preferred foot point area, the warning foot point area and the dangerous foot point area corresponding to each quadruped motion terrain based on the analysis result further comprise: Mark the area in the landing point coincidence rectangle where only the regular landing point area exists as the preferred foot point area of uphill motion, and mark the area in the landing point coincidence rectangle where only the accident landing point area exists as the dangerous foot point area of uphill motion; mark the area in the landing point coincidence rectangle where the regular landing point area and the accident landing point area coincide as the warning foot point area of uphill motion; Based on the analysis method of the uphill sensing data, analyze the downhill sensing data, the straight line sensing data and the turning sensing data respectively, and obtain the preferred foot point area, the warning foot point area and the dangerous foot point area corresponding to the downhill terrain, the straight line motion terrain and the turning terrain.

8. The large model based quadruped robot path planning method according to claim 7, characterized in that, Obtain a to-be-planned path, segment the to-be-planned path based on all quadruped motion terrains, and obtain a plurality of segmented planning paths; adjust each segmented planning path based on the preferred foothold area, warning foothold area and dangerous foothold area of the quadruped motion terrain, and record the path obtained by restoring the adjusted segmented planning path as the standard motion path of the quadruped robot, and use the standard motion path to control the motion of the robot, including: Before the quadruped robot performs motion, obtain the path of the quadruped robot when performing motion, recorded as a to-be-planned path; segment the to-be-planned path using a large model, which includes: using the large model to simulate the motion of the quadruped robot in the to-be-planned path, and recording the obtained sensor data as simulated sensor data; based on the analysis method of the large model when analyzing the motion sensor data, using the large model to obtain the uphill terrain, downhill terrain, straight motion terrain and turning terrain in the to-be-planned path from the simulated sensor data, and mark them as segmented planning paths respectively; For any one segmented planning path: based on the simulated sensor data, obtain the plan view of the path under the overhead angle when the simulated quadruped robot moves in the segmented planning path, and record it as a simulated overhead plan view; record the minimum circumscribed rectangle of the simulated overhead plan view as a simulated overhead rectangle; When the segmented planning path is an uphill terrain, use the landing point coincidence rectangle corresponding to the uphill motion as a reference to proportionally scale the simulated overhead rectangle until the landing point coincidence rectangle and the overhead simulated rectangle have the same motion direction of the quadruped robot in the overhead simulated rectangle, and the overhead simulated rectangle and the landing point coincidence rectangle can completely coincide; label the preferred foothold area, dangerous foothold area and warning foothold area that are completely the same as the landing point coincidence rectangle in the overhead simulated rectangle after proportional scaling; Restore the simulated overhead rectangle to the state before proportional scaling, and proportionally scale the preferred foothold area, dangerous foothold area and warning foothold area in the simulated overhead rectangle based on the proportion of the proportional scaling of the simulated overhead rectangle during restoration; Based on the analysis method when the segmented planning path is an uphill terrain, analyze in the same way when the segmented planning path is a downhill terrain, a straight motion terrain or a turning terrain, and obtain the corresponding simulated overhead rectangle and the preferred foothold area, dangerous foothold area and warning foothold area in the simulated overhead rectangle; Based on the position of each segmented planning path in the to-be-planned path, record the path composed of the preferred foothold areas of all segmented planning paths as the standard motion path of the quadruped robot, and perform motion along the standard motion path when the quadruped robot performs motion; when the landing point of any one mechanical foot during actual motion is in the warning foothold area of any one segmented planning path, send a path deviation warning until the landing points of all mechanical feet of the quadruped robot are in the preferred foothold area; When the landing point of any one mechanical foot during actual motion is in the dangerous foothold area of any one segmented planning path, stop the motion of the quadruped robot and report to the staff.

9. An electronic device, comprising: A computer program product comprising a processor and a memory storing computer readable instructions which, when executed by the processor, perform the steps of the method of any of claims 1-8.

10. A storage medium having stored thereon a computer program, characterized in that The computer program, when executed by a processor, performs the steps of the method of any of claims 1-8.

Images