Simulation method and device for perceptual data of foot-type robot and storage medium
By constructing a legged robot model and environment on a 3D simulation platform and adding a virtual perception model for testing, the problems of high cost and long cycle were solved, realizing a low-cost and efficient research method and promoting the rapid development of embodied intelligence technology.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-15
- Publication Date
- 2026-04-14
AI Technical Summary
The high cost of purchasing legged robot bodies with complete perception and movement functions, coupled with the high dependence of research, development and verification work on robot hardware and specific environments, results in high research thresholds and long cycles, making it difficult to promote the rapid development of embodied intelligence technology.
By constructing a 3D simulation model and experimental environment for a legged robot, adding a preset virtual perception model, controlling the target simulation model to perform actions and collecting simulation perception data, obstacle avoidance performance testing is conducted, reducing the dependence on physical robots and the actual environment.
It effectively reduces the hardware cost threshold for embodied intelligence research, shortens the research, development and verification cycle, supports the parallel operation of multiple sets of experiments, and significantly improves research efficiency.
Smart Images
Figure CN121859535A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of robot simulation technology, and more specifically, to a simulation method, device, and storage medium for sensory data of a legged robot. Background Technology
[0002] Embodied Artificial Intelligence (EAL), a cutting-edge field at the intersection of artificial intelligence and robotics, is characterized by its emphasis on intelligent agents achieving autonomous learning and evolution through dynamic interaction between their own "body" and the external environment. Its essence lies in the deep integration of perception, action, and cognition. Currently, with the rapid iteration of artificial intelligence technology and the continuous maturation of robot manufacturing processes, embodied intelligence has become a hot research direction in the global scientific research field, demonstrating broad application prospects in robotics.
[0003] However, embodied intelligence research faces two major bottlenecks. On the one hand, the cost of purchasing legged robot bodies with complete perception and movement functions is high, including hardware procurement and maintenance costs, which significantly raises the research threshold and makes it difficult for most small and medium-sized research institutions, startup teams, and individual researchers to afford. On the other hand, all research, development, and verification work is highly dependent on the robot body hardware and specific actual experimental environments, resulting in low data generation efficiency and difficulty in scenario reproduction, which significantly prolongs the research cycle and seriously restricts the development progress of embodied intelligence technology.
[0004] Therefore, developing a method that can accurately simulate the physical properties of a legged robot in a simulation platform and accurately reflect these physical characteristics in sensor data and complete the output has become the key to breaking through the current research bottleneck. Summary of the Invention
[0005] This application addresses the shortcomings of the prior art by providing a simulation method, device, and storage medium for legged robot perception data, in order to solve the problems existing in the prior art.
[0006] The technical solution adopted in the embodiments of this application is as follows: In a first aspect, embodiments of this application provide a simulation method for legged robot perception data, including: Based on the physical property data of the legged robot and the experimental environment data of the legged robot, a three-dimensional simulation model of the legged robot and a three-dimensional simulation experimental environment are constructed. A preset virtual perception model is added to the target location on the three-dimensional simulation model to obtain the target simulation model of the legged robot. The target simulation model is controlled to perform target actions in the three-dimensional simulation experimental environment, and the simulation perception data collected by the preset virtual perception model is acquired. Based on the simulated perception data, the obstacle avoidance performance of the legged robot was tested.
[0007] In one embodiment, constructing a three-dimensional simulation model and a three-dimensional simulation experimental environment for the legged robot based on its physical property data and the experimental environment data of the legged robot includes: Based on the physical property data of the legged robot, a pre-set simulation platform is used to generate symbolic representations of the inertial parameters, kinematic pair constraints, and contact geometry of the multi-rigid-body system in the legged robot. Based on the symbolic representation of the inertial parameters, kinematic pair constraints, and contact geometry of the multi-rigid-body system, the mass matrix and constraint Jacobian are constructed using the preset simulation platform, and the contact friction matrix is pre-assigned. Based on the mass matrix, the constraint Jacobian, and the contact friction matrix, motion constraint information is configured for the preset three-dimensional model of the legged robot using the preset simulation platform to obtain the three-dimensional simulation model. Based on the experimental environment data of the legged robot, the three-dimensional simulation experimental environment is constructed using the preset simulation platform.
[0008] In one embodiment, before constructing the three-dimensional simulation model and three-dimensional simulation experimental environment of the legged robot based on its physical property data and the experimental environment data of the legged robot, the method further includes: The physical attribute data, experimental environment data, and preset perception configuration data are obtained from the simulation configuration file of the legged robot. The step of adding a preset virtual perception model at the target location on the three-dimensional simulation model to obtain the target simulation model of the legged robot includes: Based on the preset perception configuration data, the preset virtual perception model is added to the target position on the three-dimensional simulation model to obtain the target simulation model of the legged robot.
[0009] In one embodiment, before adding a preset virtual perception model at the target location on the three-dimensional simulation model to obtain the target simulation model of the legged robot, the method further includes: The target position is obtained by retrieving the preset sensing installation position of the legged robot from its open-source structure file; or... The user-defined sensing installation location is determined as the target location.
[0010] In one embodiment, controlling the target simulation model to perform a target action in the three-dimensional simulation experimental environment and collecting simulation perception data collected by the preset virtual perception model includes: Based on the running sequence data of the legged robot, a motion control model of the legged robot is constructed; Based on the simulated motion data of the target action, motion control commands are generated using the motion control model. According to the motion control command, the target simulation model is controlled to perform the target action in the three-dimensional simulation experimental environment, and the simulation perception data collected by the preset virtual perception model is collected.
[0011] In one embodiment, the simulated sensing data includes: radar sensing data collected by a preset simulated radar; the method further includes: Based on the operating parameters of the preset simulation radar, construct the simulated light beam of the preset simulation radar; Based on the simulated light and the radar sensing data, a radar sensing image in the three-dimensional simulation experimental environment is generated.
[0012] In one embodiment, generating a radar perception image in the three-dimensional simulation experimental environment based on the simulated light and the radar perception data includes: Based on the simulated light rays and the radar sensing data, an intersection test is performed on each light ray with the three-dimensional simulation experimental environment to obtain the point cloud coordinates of the nearest hit for each light ray in the three-dimensional simulation experimental environment; The radar sensing image is generated based on the point cloud coordinates.
[0013] In one embodiment, the simulated perception data further includes: simulated inertial navigation data collected by a preset simulated inertial measurement unit; the step of testing the obstacle avoidance performance of the legged robot based on the simulated perception data includes: The obstacle avoidance performance of the legged robot is tested based on the radar perception data and the simulated inertial navigation data.
[0014] Secondly, embodiments of this application also provide an electronic device, including: a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to implement the simulation method for legged robot perception data described in any of the above embodiments.
[0015] Thirdly, embodiments of this application also provide a readable storage medium storing program instructions, which, when executed by a processor, implement the simulation method for legged robot perception data described in any of the above embodiments.
[0016] The beneficial effects of this application are as follows: This application provides a simulation method for legged robot perception data. By adding a preset virtual perception model at the target position on the three-dimensional simulation model to obtain the target simulation model of the legged robot, and using the target simulation model to test the robot, the hardware cost threshold of embodied intelligence research can be effectively reduced, the dependence on physical robots and actual environments can be reduced, and the research, development and verification cycle can be significantly shortened. This is of great significance for promoting the rapid development of embodied intelligence technology. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 One of the flowcharts for simulating legged robot perception data provided in the embodiments of this application; Figure 2 A second schematic flowchart illustrating the simulation method for legged robot perception data provided in this application embodiment; Figure 3 The third flowchart illustrates the simulation method for legged robot perception data provided in this application embodiment; Figure 4 The fourth flowchart illustrates the simulation method for legged robot perception data provided in this application embodiment. Figure 5 Fifth flowchart illustrating the simulation method for legged robot perception data provided in the embodiments of this application; Figure 6 A schematic diagram of the simulated light beam of the preset simulated radar provided in the embodiments of this application; Figure 7 A flowchart illustrating the simulation method for legged robot perception data provided in this application embodiment is shown in Figure 6. Figure 8 Radar sensing images provided in the embodiments of this application; Figure 9 A schematic diagram of the structure of a simulation device for legged robot perception data provided in an embodiment of this application; Figure 10This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of this application, but not all embodiments.
[0020] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.
[0021] Furthermore, the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Additionally, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0022] It should be noted that, where there is no conflict, the features in the embodiments of this application can be combined with each other.
[0023] This application provides a simulation method for legged robot perception data. This method can be generated by any electronic device with computing and processing capabilities. The electronic device can be, for example, a computer device facing the terminal or a backend server.
[0024] The following examples, in conjunction with the accompanying drawings, provide specific illustrations of the simulation method for legged robot perception data provided in this application.
[0025] Figure 1 This is one of the flowcharts illustrating the simulation method for legged robot perception data provided in the embodiments of this application, such as... Figure 1 As shown, the method includes: S101. Based on the physical property data of the legged robot and the experimental environment data of the legged robot, construct a three-dimensional simulation model of the legged robot and a three-dimensional simulation experimental environment.
[0026] Extract the physical attribute data of the legged robot and the experimental environment data of the legged robot from the robot's simulation configuration file (such as an XML-formatted structure description file) to construct a three-dimensional simulation model of the legged robot and a three-dimensional simulation experimental environment.
[0027] S102. Add a preset virtual perception model to the target location on the three-dimensional simulation model to obtain the target simulation model of the legged robot.
[0028] First, obtain the preset sensing installation position of the legged robot from the open-source structure file of the legged robot as the target position, or determine the sensing installation position set by the user as the target position.
[0029] Then, a preset virtual perception model is added to the target location on the 3D simulation model to obtain the target simulation model of the legged robot.
[0030] S103. Control the target simulation model to perform target actions in a three-dimensional simulation experimental environment, and collect simulation perception data collected by the preset virtual perception model.
[0031] The target simulation model is controlled to perform target actions in a three-dimensional simulation experimental environment, so that the target simulation model of the legged robot can realize the same standing, walking and other actions as the physical robot, and collect simulation perception data collected by the preset virtual perception model.
[0032] S104. Based on the simulation perception data, conduct obstacle avoidance performance tests on the legged robot.
[0033] In summary, this embodiment provides a simulation method for legged robot perception data. By adding a preset virtual perception model at the target position on the three-dimensional simulation model to obtain the target simulation model of the legged robot, and using the target simulation model to test the robot, the hardware cost threshold for embodied intelligence research can be effectively reduced, the dependence on physical robots and actual environments can be reduced, and the research, development and verification cycle can be significantly shortened. This is of great significance for promoting the rapid development of embodied intelligence technology.
[0034] Figure 2 This is the second flowchart illustrating the simulation method for legged robot perception data provided in the embodiments of this application. Figure 2 As shown, S101 may include: S201. Based on the physical property data of the legged robot, a pre-set simulation platform is used to generate symbolic representations of the inertial parameters, kinematic pair constraints, and contact geometry of the multi-rigid-body system in the legged robot.
[0035] Using the dynamic modeling capabilities of the pre-set simulation platform Mujoco, the physical property data of the legged robot are transformed into inertial parameters (mass, moment of inertia), kinematic pair constraints (joint range of motion, axial direction), and contact geometry parameters (collision surface shape, friction coefficient) of a multi-rigid-body system.
[0036] S202. Based on the symbolic representation of the inertial parameters, kinematic pair constraints, and contact geometry of the multi-rigid-body system, a pre-set simulation platform is used to construct the mass matrix and constraint Jacobian, and the contact friction matrix is pre-assigned.
[0037] Based on the symbolic representation of the multi-rigid-body system, Mujoco automatically calculates and constructs the mass matrix in generalized coordinates, while generating the constraint Jacobian matrix (used to describe the relationship between joint constraints and motion), and pre-assigns the contact friction matrix (for example, setting the ground friction coefficient to 0.8, which conforms to the physical characteristics of the real ground).
[0038] S203. Based on the mass matrix, constraint Jacobian, and contact friction matrix, the motion constraint information is configured for the preset three-dimensional model of the legged robot using a preset simulation platform to obtain the three-dimensional simulation model.
[0039] Mujoco uses an improved Projected Gauss-Seidel solver to map the mass matrix, constraint Jacobian matrix, and contact friction matrix into motion constraint information for the robot. It configures a preset 3D model (such as a robot mesh model provided by the manufacturer) to generate a 3D simulation model that can move according to physical laws. For example, it adds floating_base_joints (free joints) to the robot to ensure that it can achieve full-body movements such as standing and walking.
[0040] S204. Based on the experimental environment data of the legged robot, a three-dimensional simulation experimental environment is constructed using a pre-set simulation platform.
[0041] The experimental environment data of the legged robot may include, for example, the dimensions of the environmental boundary, the coefficient of friction of the ground, the location and geometry of obstacles (e.g., a cube obstacle with a side length of 0.5m and coordinates (2,1,0)).
[0042] Based on experimental environment data, a virtual environment was built in Mujoco, including elements such as ground and obstacles. The geometry of the environment was defined by Mujoco's geom tags, such as using the type="mesh" tag to load the ground mesh and the type="box" tag to create obstacles, ultimately forming a three-dimensional simulation experimental environment consistent with the real experimental scene.
[0043] Figure 3 The third flowchart illustrates the simulation method for legged robot perception data provided in this application embodiment. Figure 3 As shown, before performing step S101, the method of this application may further include: S301. Obtain physical attribute data, experimental environment data, and preset perception configuration data from the simulation configuration file of the legged robot.
[0044] Data is preferentially obtained from the open-source simulation configuration files provided by the legged robot manufacturer. If it is an independently modified robot, the R&D personnel determine the key parameters based on the modification design.
[0045] S102, which describes adding a preset virtual perception model at the target location on the three-dimensional simulation model to obtain the target simulation model of the legged robot, may include: S302. Based on the preset perception configuration data, add a preset virtual perception model at the target position on the three-dimensional simulation model to obtain the target simulation model of the legged robot.
[0046] Taking a radar as an example, the preset virtual perception model includes the radar's installation location data. Using Mujoco's MJCF (Mujoco XML Configuration Format) technology, the simulation configuration file is modified to declare the element-attribute tree. The radar's installation location data is added to the simulation configuration file, and a preset virtual perception model can be added to the target location on the 3D simulation model to obtain the target simulation model of the legged robot.
[0047] Figure 4 The fourth flowchart illustrates the simulation method for legged robot perception data provided in this application embodiment. Figure 4 As shown, the control target simulation model described in S103 executes target actions in a three-dimensional simulation experimental environment and collects simulation perception data collected by a preset virtual perception model, which may include: S401. Based on the running sequence data of the legged robot, construct a motion control model for the legged robot.
[0048] A motion control model capable of outputting joint control commands is trained by using reinforcement learning algorithms and robot motion sequence data (such as joint angles when standing and gait cycles when walking) as training samples.
[0049] Alternatively, if a motion control model is provided by the robot manufacturer, it can be directly imported into Mujoco.
[0050] S402. Based on the simulated motion data of the target action, a motion control model is used to generate motion control commands.
[0051] The motion control model outputs control commands for each joint motor by observing the simulated motion data of the robot in Mujoco (such as the current angle of the joints and the position of the center of mass).
[0052] S403. According to the motion control command, control the target simulation model to perform target actions in the three-dimensional simulation experimental environment, and collect simulation perception data collected by the preset virtual perception model.
[0053] Based on the test requirements (such as the obstacle avoidance test requiring "straight walking + turning"), set the simulated motion data of the target action (such as walking speed of 0.5m / s and turning angle of 90°).
[0054] The motion control model generates motion control commands based on this data, driving the target simulation model to perform target actions in a three-dimensional simulation experimental environment. During the action, the physical states (such as joint torques and center of mass acceleration) of the robot and the preset virtual perception model are consistent with those of the physical robot. In this process, the simulation perception data collected by the preset virtual perception model is acquired in real time.
[0055] In one embodiment, the preset virtual sensing model is a preset simulated radar, and the simulated sensing data includes radar sensing data collected by the preset simulated radar. Figure 5 The fifth flowchart illustrates the simulation method for legged robot perception data provided in this application embodiment. Figure 5 As shown, in this case, the method of this application further includes: S501. Based on the preset operating parameters of the simulated radar, construct the simulated light beam of the preset simulated radar.
[0056] The preset operating parameters of the simulated radar may include, for example, the radar's physical parameters (detection range, number of beams, scanning frequency), internal parameter files (focal length, distortion coefficient), and the IMU's measurement range (acceleration ±16g, angular velocity ±2000° / s), etc.
[0057] Based on the above operating parameters, a simulated light beam for a preset simulated radar is constructed, such as... Figure 6 As shown in the image, the propagation path of the simulated light rays, their intersections with obstacles, and the distribution of point clouds (e.g., the point cloud density in obstacle areas is higher than in open areas).
[0058] S502. Based on the simulated light and radar sensing data, generate a radar sensing image in the three-dimensional simulation experimental environment.
[0059] Based on simulated light and radar sensing data, a radar sensing image in a three-dimensional simulation experimental environment can be generated. Specifically, Figure 7 The sixth flowchart illustrates the simulation method for legged robot perception data provided in this application embodiment. Figure 7As shown, S502 may include: S601. Based on the simulated light rays and radar sensing data, perform an intersection test between each light ray and the three-dimensional simulation experimental environment to obtain the coordinates of the point cloud where each light ray is most recently hit in the three-dimensional simulation experimental environment.
[0060] Multithreading is launched in the GPU, with each thread responsible for the intersection test of one simulated ray. Each thread reads the starting coordinates (radar installation location) and direction vector of a single ray and traverses the environmental partitions along its propagation path.
[0061] For triangular faces within a partition, the intersecting ray is calculated using a ray-facet intersection formula (such as the Moller–Trumbore algorithm). The spatial coordinates of the intersection point and the facet material are recorded (for subsequent point cloud reflectivity assignment). If a ray intersects multiple faces, only the coordinates of the intersection point closest to the radar are retained (excluding invalid points at a distance to simulate the "near-point priority detection" characteristic of real radar).
[0062] After all threads have completed the test, the coordinates (x, y, z) of the nearest hit point of each ray are collected into a point cloud coordinate set and sent back to the CPU memory to form a single frame of radar point cloud data. For example, the coordinates of the nearest hit point of a certain ray (vertical angle +5°, horizontal angle 90°) are (3, 1, 0.4), which corresponds to the front side of the cube obstacle.
[0063] S602. Generate radar sensing images based on point cloud coordinates.
[0064] The world coordinate system (x, y, z) of the point cloud is converted to the radar coordinate system (with the radar installation location as the origin, x-axis forward, y-axis left, and z-axis upward), ensuring that the image viewpoint matches the actual radar detection viewpoint. Then, based on Mujoco, the processed point cloud coordinates are converted into a radar sensing image, such as... Figure 8 As shown.
[0065] In one embodiment, the simulated perception data further includes simulated inertial navigation data collected by a preset simulated inertial measurement unit (IMU). Step S104, which involves testing the obstacle avoidance performance of the legged robot based on the simulated perception data, may include: testing the obstacle avoidance performance of the legged robot based on radar perception data and simulated inertial navigation data. The specific steps are as follows: In a 3D simulation environment, obstacles (such as a cylinder with a diameter of 1m, located 3m in front of the robot's walking path) are set, and the target action is set as "the robot walks in a straight line at a speed of 0.5m / s and turns to avoid the obstacle after encountering it". The radar perception data and the simulated inertial navigation data are synchronized in time. The position and outline of the obstacle are identified by the radar data, and the current position and attitude of the robot are calculated by combining the simulated inertial navigation data (such as obtaining the robot's displacement by integrating the acceleration of the IMU and obtaining the turning angle by integrating the angular velocity).
[0066] The test determines whether the robot can accurately identify obstacles based on simulation perception data and adjust its trajectory to avoid them. If the robot successfully starts turning 1m in front of the obstacle and avoids it without collision, the obstacle avoidance performance is deemed qualified. If the robot fails to identify the obstacle or a collision occurs, the motion control model or perception parameters are optimized until the test requirements are met.
[0067] In summary, this application provides a simulation method for legged robot perception data, which has the following advantages: 1. High physical accuracy: Based on Mujoco's dynamics solver, the simulation data (such as joint torques and point cloud coordinates) deviates from the physical characteristics of the real robot by less than 5%, which can replace physical hardware for preliminary algorithm verification and effectively reduce the hardware cost threshold for embodied intelligence research.
[0068] 2. Low cost and high efficiency: There is no need to purchase expensive physical robots. R&D personnel can carry out experiments through ordinary workstations, and multiple sets of experiments can be run in parallel, shortening the R&D cycle by more than 40%.
[0069] 3. Flexible and scalable: Supports simulation of different types of legged robots (such as quadruped and bipedal), and allows for quick replacement of perception model parameters by modifying XML files (such as replacing 16-line radar with 64-line radar).
[0070] The following will continue to explain the apparatus, device, and storage medium for implementing the simulation method of legged robot perception data provided in any of the above embodiments of this application. The specific implementation process and the resulting technical effects are the same as those in the corresponding method embodiments. For the sake of brevity, the parts not mentioned in the following embodiments can be referred to the corresponding content in the method embodiments.
[0071] Figure 9 This is a schematic diagram of the structure of the simulation device for legged robot perception data provided in the embodiments of this application, as shown below. Figure 9 As shown, this application provides a simulation device for legged robot perception data, comprising: The construction module 10 is used to construct a three-dimensional simulation model of the legged robot and a three-dimensional simulation experimental environment based on the physical attribute data of the legged robot and the experimental environment data of the legged robot.
[0072] The acquisition module 20 is used to add a preset virtual perception model at the target position on the three-dimensional simulation model to obtain the target simulation model of the legged robot.
[0073] The control module 30 is used to control the target simulation model to perform target actions in the three-dimensional simulation experimental environment and to collect simulation perception data collected by the preset virtual perception model.
[0074] The test module 40 is used to test the obstacle avoidance performance of the legged robot based on the simulated perception data.
[0075] Optionally, the construction module 10 is used to generate symbolic representations of the inertial parameters, kinematic pair constraints, and contact geometry of the multi-rigid-body system in the legged robot based on the physical property data of the legged robot using a preset simulation platform; construct a mass matrix and constraint Jacobian based on the symbolic representations of the inertial parameters, kinematic pair constraints, and contact geometry of the multi-rigid-body system using the preset simulation platform, and pre-allocate a contact friction matrix; configure motion constraint information for a preset three-dimensional model of the legged robot using the preset simulation platform based on the mass matrix, the constraint Jacobian, and the contact friction matrix, thereby obtaining the three-dimensional simulation model; and construct the three-dimensional simulation experimental environment based on the experimental environment data of the legged robot using the preset simulation platform.
[0076] Optionally, the acquisition module is further configured to acquire the physical attribute data, the experimental environment data, and the preset perception configuration data from the simulation configuration file of the legged robot; and add the preset virtual perception model at the target position on the three-dimensional simulation model according to the preset perception configuration data to obtain the target simulation model of the legged robot.
[0077] Optionally, the acquisition module is further configured to acquire the preset sensing installation position of the legged robot from the open-source structure file of the legged robot as the target position; or, determine the sensing installation position set by the user as the target position.
[0078] Optionally, the control module 30 is further configured to construct a motion control model of the legged robot based on the running sequence data of the legged robot; generate motion control commands using the motion control model based on the simulated motion data of the target action; control the target simulation model to perform the target action in the three-dimensional simulation experimental environment according to the motion control commands, and collect the simulation perception data collected by the preset virtual perception model.
[0079] Optionally, the simulation perception data includes: radar perception data collected by a preset simulation radar; the construction module 10 is further configured to construct the simulation light beam of the preset simulation radar according to the operating parameters of the preset simulation radar; and generate a radar perception image in the three-dimensional simulation experimental environment according to the simulation light beam and the radar perception data.
[0080] Optionally, the construction module 10 is further configured to perform an intersection test between each ray and the three-dimensional simulation experimental environment based on the simulated ray and the radar sensing data, to obtain the point cloud coordinates of the nearest hit of each ray in the three-dimensional simulation experimental environment; and to generate the radar sensing image based on the point cloud coordinates.
[0081] Optionally, the simulated perception data further includes: simulated inertial navigation data collected by a preset simulated inertial measurement unit; the test module 40 is also used to perform obstacle avoidance performance testing on the legged robot based on the radar perception data and the simulated inertial navigation data.
[0082] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.
[0083] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more microprocessors, or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).
[0084] Figure 10 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application, such as... Figure 10 As shown, this application also provides an electronic device, including a processor 100, a storage medium 200 and a bus 300. The storage medium stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to implement the simulation method for legged robot perception data described in any of the above embodiments.
[0085] This application also provides a readable storage medium storing program instructions, which, when executed by a processor, implement the simulation method for legged robot perception data described in any of the above embodiments.
[0086] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0087] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0088] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units.
[0089] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0090] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A simulation method for sensory data of a legged robot, characterized in that, include: Based on the physical property data of the legged robot and the experimental environment data of the legged robot, a three-dimensional simulation model of the legged robot and a three-dimensional simulation experimental environment are constructed. A preset virtual perception model is added to the target location on the three-dimensional simulation model to obtain the target simulation model of the legged robot. The target simulation model is controlled to perform target actions in the three-dimensional simulation experimental environment, and the simulation perception data collected by the preset virtual perception model is acquired. Based on the simulated perception data, the obstacle avoidance performance of the legged robot was tested.
2. The method according to claim 1, characterized in that, The process of constructing a three-dimensional simulation model and a three-dimensional simulation experimental environment for the legged robot based on its physical property data and the experimental environment data includes: Based on the physical property data of the legged robot, a pre-set simulation platform is used to generate symbolic representations of the inertial parameters, kinematic pair constraints, and contact geometry of the multi-rigid-body system in the legged robot. Based on the symbolic representation of the inertial parameters, kinematic pair constraints, and contact geometry of the multi-rigid-body system, the mass matrix and constraint Jacobian are constructed using the preset simulation platform, and the contact friction matrix is pre-assigned. Based on the mass matrix, the constraint Jacobian, and the contact friction matrix, motion constraint information is configured for the preset three-dimensional model of the legged robot using the preset simulation platform to obtain the three-dimensional simulation model. Based on the experimental environment data of the legged robot, the three-dimensional simulation experimental environment is constructed using the preset simulation platform.
3. The method according to claim 1, characterized in that, Before constructing the three-dimensional simulation model and three-dimensional simulation experimental environment of the legged robot based on its physical property data and the experimental environment data of the legged robot, the method further includes: The physical attribute data, experimental environment data, and preset perception configuration data are obtained from the simulation configuration file of the legged robot. The step of adding a preset virtual perception model at the target location on the three-dimensional simulation model to obtain the target simulation model of the legged robot includes: Based on the preset perception configuration data, the preset virtual perception model is added to the target position on the three-dimensional simulation model to obtain the target simulation model of the legged robot.
4. The method according to claim 1, characterized in that, Before adding a preset virtual perception model at the target location on the three-dimensional simulation model to obtain the target simulation model of the legged robot, the method further includes: The target position is obtained by retrieving the preset sensing installation position of the legged robot from its open-source structure file; or... The user-defined sensing installation location is determined as the target location.
5. The method according to claim 1, characterized in that, The control of the target simulation model to perform target actions in the three-dimensional simulation experimental environment and the collection of simulation perception data acquired by the preset virtual perception model include: Based on the running sequence data of the legged robot, a motion control model of the legged robot is constructed; Based on the simulated motion data of the target action, motion control commands are generated using the motion control model. According to the motion control command, the target simulation model is controlled to perform the target action in the three-dimensional simulation experimental environment, and the simulation perception data collected by the preset virtual perception model is collected.
6. The method according to claim 1, characterized in that, The simulated sensing data includes: radar sensing data collected by a preset simulated radar; the method further includes: Based on the operating parameters of the preset simulation radar, construct the simulated light beam of the preset simulation radar; Based on the simulated light and the radar sensing data, a radar sensing image in the three-dimensional simulation experimental environment is generated.
7. The method according to claim 6, characterized in that, The step of generating a radar perception image in the three-dimensional simulation experimental environment based on the simulated light and the radar perception data includes: Based on the simulated light rays and the radar sensing data, an intersection test is performed on each light ray with the three-dimensional simulation experimental environment to obtain the point cloud coordinates of the nearest hit for each light ray in the three-dimensional simulation experimental environment; The radar sensing image is generated based on the point cloud coordinates.
8. The method according to claim 6, characterized in that, The simulated perception data further includes: simulated inertial navigation data collected by a preset simulated inertial measurement unit; the step of testing the obstacle avoidance performance of the legged robot based on the simulated perception data includes: The obstacle avoidance performance of the legged robot is tested based on the radar perception data and the simulated inertial navigation data.
9. An electronic device, characterized in that, include: The device includes a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to implement the simulation method for legged robot perception data as described in any one of claims 1 to 8.
10. A readable storage medium, characterized in that, The readable storage medium stores program instructions, which, when executed by a processor, implement the simulation method for legged robot perception data as described in any one of claims 1 to 8.