Interaction test method and device for gait simulation of humanoid robot

Abstract

The invention discloses an interaction test method for gait simulation of a humanoid robot, which is applied to a test platform. Comprising the following steps: controlling a current gait simulation model corresponding to a virtual robot to perform a gait simulation test in a virtual test environment in a current preset time period; the virtual test data is sent to the physical robot; controlling the physical robot to execute corresponding gait operation according to the virtual test data in the actual test environment, and sending the actual test data to the virtual robot; updating the current gait simulation model based on the virtual test data and the real test data; and continuing to perform the gait simulation test based on the updated gait simulation model, and ending the updating operation until the gait simulation test is ended, so as to generate a complete gait scheme. Therefore, by constructing a closed-loop interaction mechanism between the virtual robot and the physical robot, dynamic adjustment of the gait simulation model is achieved, the gait of the physical robot is restored, and the stability and precision of the gait of the physical robot are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of humanoid robots, and in particular relates to an interactive testing method and device for gait simulation of a humanoid robot. Background Art

[0002] As humanoid robot technology continues to advance, how to efficiently and accurately evaluate its gait performance has become a key point in improving the movement capabilities of humanoid robots. Currently, most gait testing methods rely heavily on physical testing environments and manually set gait simulation scenarios. However, this traditional testing method has obvious shortcomings. On the one hand, the physical space size of the test site limits the test scope and cannot fully simulate complex and diverse real-world scenarios; on the other hand, building and maintaining a physical test environment often requires high costs, and the testing process is time-consuming. These factors combined make it difficult for existing testing methods to comprehensively and accurately evaluate the gait performance of humanoid robots in complex environments.

[0003] The emergence of virtual simulation technology has opened up a new, more flexible and efficient path for humanoid robot gait testing. Virtual simulation technology can break free from the constraints of physical space, rapidly construct a variety of complex scenarios, and greatly improve testing efficiency. However, it cannot be ignored that there are still difficult technical challenges to overcome in terms of data transmission and synchronization of motion feedback between virtual and physical robots. In view of this, establishing a real-time data exchange and feedback mechanism between virtual and physical robots to achieve accurate evaluation of humanoid robot gait performance has become a key direction of current technical research in this field. Summary of the Invention

[0004] In response to the above-mentioned problems existing in the prior art, an embodiment of the present invention provides an interactive testing method and device for humanoid robot gait simulation; this method can solve technical problems in the prior art such as the physical testing environment being insufficiently flexible and difficult to construct, and the robot gait simulation being insufficiently accurate.

[0005] According to a first aspect of an embodiment of the present invention, an interactive testing method for gait simulation of a humanoid robot is provided, which is applied to a testing platform; the testing platform includes a physical robot and a virtual robot; the physical robot is communicatively connected to the virtual robot; the method includes: controlling a current gait simulation model corresponding to the virtual robot to perform a gait simulation test in a virtual testing environment in a current preset time period to generate current virtual test data; and sending the current virtual test data to the physical robot; controlling the physical robot to perform corresponding gait operations according to the current virtual test data in an actual testing environment corresponding to the virtual testing environment to generate current real test data; and sending the current real test data to the virtual robot; based on the current virtual test data and the current real test data, performing an update operation on the current gait simulation model to generate an updated gait simulation model; using the updated gait simulation model as the next gait simulation model corresponding to the next preset time period adjacent to the current preset time period, and continuing to perform gait simulation testing in the next preset time period, until the gait simulation test process is completed and the update operation of the gait simulation model is terminated to generate a complete gait scheme corresponding to the virtual testing environment.

[0006] Optionally, the method further includes: constructing a virtual test environment according to the gait test task; constructing a gait test platform; wherein the gait test platform is used to indicate a platform capable of regulating the test parameters of the test environment; the test parameters include at least: test speed, and / or ground friction coefficient, and / or test slope, and / or right-angle turn; based on the virtual test environment, the gait test platform is controlled in a coordinated manner to generate an actual test environment corresponding to the virtual test environment.

[0007] Optionally, the method further includes: using kinematics and dynamics modeling tools to construct a virtual robot motion model according to the specific structure of the physical robot; configuring a control algorithm corresponding to the gait test task for the virtual robot motion model to obtain a gait simulation model.

[0008] Optionally, the method further includes: obtaining several complete gait schemes based on the complete gait scheme generated by each gait simulation test in several gait simulation tests; wherein each of the complete gait schemes includes quasi-virtual test data formed by at least two groups of virtual test data and quasi-real test data formed by at least two groups of real test data; for any of the several complete gait schemes: determining the quasi-gait deviation of the complete gait scheme based on the quasi-virtual test data and quasi-real test data corresponding to the complete gait scheme, and generating a decision matrix; wherein the quasi-gait deviation includes at least quasi-gait joint point position deviation, quasi-joint angle deviation, quasi-zero torque point deviation, quasi-center of mass trajectory deviation, and minimum quasi-energy consumption; the quasi-gait deviation in the decision matrix is ​​converted into the quasi-gait deviation. The quasi-gait deviation is normalized to generate a standard matrix formed by standardized values; a corresponding weight is applied to each standardized value in the standard matrix to generate an update matrix; a positive ideal solution and a negative ideal solution are calculated according to the update matrix; the distance between each of the update matrices and the positive ideal solution and the negative ideal solution is calculated to obtain a corresponding first distance and a second distance; the relative closeness of the complete gait scheme is determined based on the first distance and the second distance; based on the relative closeness corresponding to each of the complete gait schemes, a plurality of relative closenesses are obtained; and the complete gait scheme with the largest relative closeness is selected from the plurality of relative closenesses as the optimal gait scheme for the virtual robot to perform a gait simulation test on the virtual test environment.

[0009] Optionally, based on the current virtual test data and the current real test data, an update operation is performed on the current gait simulation model to generate an updated gait simulation model; including: based on the current virtual test data and the current real test data, determining the current gait deviation corresponding to the virtual robot; wherein the current gait deviation at least includes: the current gait joint point position deviation, the current joint angle deviation, the current zero torque point deviation, and the current center of mass trajectory deviation; based on the current gait deviation and the current real test data, gait error correction is performed on the virtual robot to generate the starting gait of the virtual robot in the next preset time period; based on the current gait deviation and the starting gait, the sub-gait scheme corresponding to the virtual robot in the next preset time period is predicted; based on the sub-gait scheme corresponding to the next time period, an update operation is performed on the current gait simulation model to generate an updated gait simulation model.

[0010] Optionally, based on the current gait deviation and the starting gait, the sub-gait scheme corresponding to the virtual robot in the next preset time period is predicted; including: based on the current gait joint point position deviation, the current joint angle deviation, the current zero torque point deviation, and the current center of mass trajectory deviation, determining the optimization objective function of the current gait simulation model; according to the starting gait, the joint angle of the virtual robot is optimized and controlled based on PID gain tuning to generate an optimized joint angle; based on the optimization objective function and the optimized joint angle, determining the constraint energy consumption corresponding to the reward function; and determining the gait scheme corresponding to the minimum constraint energy consumption as the sub-gait scheme corresponding to the virtual robot in the next preset time period.

[0011] Optionally, the current preset time period includes several acquisition time points; the current virtual test data includes several groups of virtual test data; the current real test data includes several groups of real test data; each of the acquisition time points corresponds to a group of virtual test data and a group of real test data; the virtual test data includes several first parameters; the real test data includes several second parameters; for any acquisition time point within the current preset time period: based on the first parameter and the second parameter corresponding to the acquisition time point, determine the parameter deviation corresponding to the acquisition time point; based on the parameter deviation corresponding to each of the acquisition time points, determine the average value of the parameter deviation corresponding to the current preset time period; based on the parameter deviation corresponding to each of the acquisition time points, the average value of the parameter deviation, and the number of acquisition time points in the current preset time period, generate the variance of the parameter deviation; determine the variance of the parameter deviation as an indicator for evaluating the gait performance of the virtual robot.

[0012] According to the second aspect of the embodiment of the present invention, there is also provided an interactive test device for gait simulation of a humanoid robot, which is characterized by being applied to a test platform; the test platform includes a physical robot and a virtual robot; the physical robot is communicatively connected to the virtual robot; a gait simulation test module is used to control the current gait simulation model corresponding to the virtual robot to perform a gait simulation test in a virtual test environment in a current preset time period to generate current virtual test data; and the current virtual test data is sent to the physical robot; a gait synchronization operation module is used to control the physical robot to perform a gait simulation test in an actual test environment corresponding to the virtual test environment according to the current virtual test data. The data performs the corresponding gait operation to generate the current real test data; and sends the current real test data to the virtual robot; a model update operation module is used to perform an update operation on the current gait simulation model based on the current virtual test data and the current real test data to generate an updated gait simulation model; a generation module is used to use the updated gait simulation model as the next gait simulation model corresponding to the next preset time period adjacent to the current preset time period, and continue to perform gait simulation test in the next preset time period, and end the update operation of the gait simulation model until the gait simulation test process is completed, and generate a complete gait plan corresponding to the virtual test environment.

[0013] According to a third aspect of an embodiment of the present invention, an electronic device is further provided, comprising: a processor; a memory for storing instructions executable by the processor; and the processor for reading the executable instructions from the memory and executing the instructions to implement the method described in the first aspect.

[0014] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is further provided, on which a computer program is stored. When the program is executed by a processor, the method described in the first aspect is implemented.

[0015] An embodiment of the present invention provides an interactive testing method for gait simulation of a humanoid robot, which is applied to a testing platform; the testing platform includes a physical robot and a virtual robot; the physical robot is communicatively connected to the virtual robot; the method includes: first, controlling a current gait simulation model corresponding to the virtual robot to perform a gait simulation test in a virtual testing environment in a current preset time period to generate current virtual test data; and sending the current virtual test data to the physical robot; second, controlling the physical robot to perform corresponding gait operations according to the current virtual test data in an actual testing environment corresponding to the virtual testing environment to generate current real test data; and sending the current real test data to the virtual robot; thereafter, based on the current virtual test data and the current real test data, performing an update operation on the current gait simulation model to generate an updated gait simulation model; finally, using the updated gait simulation model as a next gait simulation model corresponding to a next preset time period adjacent to the current preset time period, continuing to perform gait simulation testing in the next preset time period, and ending the update operation of the gait simulation model after the gait simulation test process is completed to generate a complete gait scheme corresponding to the virtual testing environment. By establishing a closed-loop interaction mechanism between virtual and physical robots, the method in this embodiment not only enables dynamic adjustment and continuous optimization of the gait simulation model, but also ensures that the gait simulation model closely replicates the actual performance of the physical robot. This not only improves the stability and accuracy of the physical robot's gait, but also significantly enhances the physical robot's adaptability and performance in complex scenarios. This addresses the technical issue of inaccurate robot gait simulation in existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Hereinafter, some specific embodiments of the present invention will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the accompanying drawings:

[0017] Figure 1 A schematic flow chart of an interactive testing method for gait simulation of a humanoid robot provided by one embodiment of the present invention;

[0018] Figure 2 A schematic diagram of a process for determining an optimal gait solution for a virtual robot in a virtual test environment according to an embodiment of the present invention;

[0019] Figure 3 A schematic diagram of an interactive testing system provided by an embodiment of the present invention;

[0020] Figure 4 A top view and a side view of a test platform provided in one embodiment of the present invention;

[0021] Figure 5 A schematic structural diagram of an interactive testing device for gait simulation of a humanoid robot provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purposes, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0023] like Figure 1 FIG. 1 is a flow chart of an interactive testing method for gait simulation of a humanoid robot provided by an embodiment of the present invention.

[0024] An interactive testing method for gait simulation of a humanoid robot is applied to a testing platform; the testing platform includes a physical robot and a virtual robot; the physical robot is communicatively connected with the virtual robot; and the method includes at least the following steps:

[0025] S101, controlling the current gait simulation model corresponding to the virtual robot to perform a gait simulation test in a virtual test environment within a current preset time period to generate current virtual test data; and sending the current virtual test data to the physical robot;

[0026] S102, controlling the physical robot to perform corresponding gait operations according to the current virtual test data in an actual test environment corresponding to the virtual test environment, generating current real test data; and sending the current real test data to the virtual robot;

[0027] S103, performing an update operation on the current gait simulation model based on the current virtual test data and the current real test data to generate an updated gait simulation model;

[0028] S104, using the updated gait simulation model as the next gait simulation model corresponding to the next preset time period adjacent to the current preset time period, and continuing to perform gait simulation testing in the next preset time period, until the gait simulation test process is completed, the update operation of the gait simulation model is terminated, and a complete gait plan corresponding to the virtual test environment is generated.

[0029] In S101 , the gait of the virtual robot is simulated.

[0030] Using kinematic and dynamic modeling tools, a virtual robot motion model is constructed based on the specific structure of the physical robot. A control algorithm corresponding to the gait test task is configured for the virtual robot motion model to obtain a gait simulation model. The gait simulation model is used to simulate the gait behavior of the virtual robot in various scenarios.

[0031] A virtual test environment is constructed, and the virtual robot's current gait simulation model is controlled to perform gait simulation tests within the virtual test environment. Gait motion data (such as joint angles, stride length, and gait cycle) is generated, and dynamic environmental data (such as ground contact force, acceleration, and velocity) is collected through virtual sensors. The motion data and dynamic environmental data are used as the current virtual test data, and the current virtual test data is transmitted in real time to the physical robot for execution and verification.

[0032] For example: S1, create a virtual test environment with various environmental conditions through virtual simulation software (such as Gazebo, V-REP, etc.). These environments include various ground types (such as smooth ground, rough ground, sand, grass, etc.), different slope ranges (such as 0° to 30° ramps), and simulated weather changes (such as rainy and snowy weather). The various elements of these virtual test environments can be flexibly adjusted through parameterized configuration, so that the virtual test environment is highly configurable and supports the simulation and testing of various gaits (such as normal walking, accelerated walking, turning, running, etc.); thereby ensuring that the performance of the virtual robot in various walking modes can be fully evaluated. To this end, the construction of the virtual test scene in this embodiment can not only meet the diverse testing needs, but also ensure a high degree of test accuracy. These virtual test environment settings can help study the movement performance of humanoid robots under various extreme conditions, and can provide data support for subsequent optimization.

[0033] S2, by introducing the kinematic and dynamic models of the virtual robot, accurately simulates the motion of the physical robot's joints, limbs, and skeleton, including key parameters such as joint angles, gait cycle, and step variation. Based on the specific requirements of the virtual robot's tasks, appropriate control algorithms, such as PID control, dynamic programming, and optimal control algorithms, are selected to optimize the smoothness of gait generation and gait transitions, thereby improving the motion performance and stability of the virtual robot's gait simulation model under different test environments and test tasks.

[0034] S3 deploys a variety of virtual sensors in a virtual test environment, including ground contact force sensors, acceleration sensors, velocity sensors, etc.; virtual sensors are used to monitor the environmental response and motion state of the virtual robot in real time during gait simulation, which is conducive to the comprehensive analysis and evaluation of the humanoid robot's motion performance and adaptability in the virtual environment.

[0035] S4: The collected sensor data is transmitted to the physical robot via a real-time data transmission system (e.g., wireless communication protocol, data bus, etc.). This process ensures gait synchronization between the virtual robot and the physical robot, ensuring that the virtual test environment guides the physical robot.

[0036] Therefore, this embodiment uses high-precision modeling tools to construct a virtual test scenario, covering elements such as different ground surfaces, slopes, obstacles, and weather conditions. Combining kinematic and dynamic models, it simulates parameters such as the humanoid robot's joint angles and gait cycle, and optimizes gait smoothness through control algorithms to improve the virtual robot's motion stability. A gait generation algorithm is used to simulate various gait types and comprehensively evaluate the virtual robot's performance. Virtual sensors are deployed to collect data such as ground contact force, acceleration, and velocity in real time, providing precise support for gait optimization. A real-time data transmission system is used to synchronize virtual data with the physical robot, enabling efficient guidance and cost optimization of the physical robot.

[0037] In S102 , the physical robot action feedback is performed.

[0038] The physical robot performs corresponding gait movements in the actual test environment based on the gait movement data provided by the virtual environment. The gait movement data is collected and fed back through its own sensors (such as joint sensors and force sensors). This data is then transmitted to the virtual environment via real-time communication to adjust and optimize the current gait simulation model in real time.

[0039] For example, in S1, the physical robot performs gait motions in the actual test environment based on gait motion data generated in the virtual test environment. The physical robot must closely replicate the virtual robot's gait simulation performance, taking into account factors such as actual hardware limitations and sensor feedback. By accurately mimicking the virtual robot's performance in actions like walking, running, and turning, the physical robot can accurately execute and provide real-time motion feedback.

[0040] In S2, the physical robot monitors dynamic environmental data during gait execution in real time using a variety of physical sensors (such as joint sensors, force sensors, and accelerometers). These sensors monitor key metrics such as joint angles, ground contact force, and gait stability, and analyze the robot's performance in different environmental conditions in real time, including gait instability and gait errors, to identify potential problems.

[0041] In step S3, the physical robot transmits the collected gait motion data and dynamic environment data back to the virtual environment via a high-speed communication protocol as real test data. This data is fed back to the virtual robot, ensuring effective interaction and adjustment between the virtual and physical robots, supporting real-time optimization of the gait simulation model within the virtual test environment.

[0042] In step S4, the virtual robot adjusts the gait simulation model based on real-world test data fed back by the physical robot. By adjusting the gait simulation model's motion parameters, gait cycle, joint angles, and other parameters, the model's performance is optimized to more accurately align with the actual movements of the physical robot. This data-driven gait optimization ensures the gait simulation model's adaptability and stability across various test scenarios.

[0043] Therefore, this embodiment ensures a high degree of consistency between the gait simulation model of the virtual robot and the physical robot through real-time interaction and feedback between the physical robot and the virtual environment, thereby improving the accuracy and reliability of the gait simulation. The data-driven optimization mechanism enables the physical robot to adjust its gait in real time during execution, enhancing its adaptability and flexibility in complex environments. Through sensor monitoring and data analysis, the robot can promptly identify potential problems in its gait and optimize them, thereby improving stability and motion performance. In addition, the real-time feedback system between the virtual environment and the physical robot reduces the trial-and-error costs and risks in traditional physical testing, improves test efficiency, reduces R&D cycles and costs, and ensures the efficient operation and stability of the physical robot in actual applications.

[0044] In S103 and S104 , the virtual robot synchronizes its gait with that of the physical robot.

[0045] The virtual robot uses the current real-world test data fed back by the physical robot to update its current gait simulation model, generating an updated gait simulation model. Based on the updated gait simulation model, the virtual test environment undergoes gait simulation testing at the next preset time interval. The gait simulation model update continues until the gait simulation test is complete, generating a complete gait solution for the virtual test environment. This forms a closed-loop interaction mechanism between the virtual robot and the physical robot, ensuring the real-time, robust, and high-precision nature of the virtual robot's gait testing.

[0046] For example, S1 collects and processes the current real-world test data provided by the physical robot. Based on this real-world and virtual test data, it performs a multi-dimensional analysis of the virtual robot's gait stability, movement completion, joint angles, and other parameters to identify potential issues in the virtual robot's execution. It also analyzes the gait error between the physical and virtual robots, assesses the virtual robot's gait accuracy and stability, and identifies any maladaptive behavior.

[0047] S2 adjusts the gait simulation model based on the performance data provided by the physical robot. For example, the virtual robot's gait parameters, such as walking trajectory, joint angles, and stride length, are optimized to better simulate the actual performance of the physical robot. Machine learning methods or control optimization algorithms are used to further train and calibrate the gait simulation model to adapt to more complex scenarios.

[0048] In step S3, the virtual test data generated by the updated gait simulation model is transmitted back to the physical robot, ensuring that the robot can continue to perform the corresponding movements according to the adjusted virtual test data. This closed-loop data transmission and real-time feedback ensure continuous gait synchronization between the virtual and physical robots.

[0049] S4 forms a closed-loop interaction mechanism by continuously transferring and optimizing data between the virtual and physical robots. The virtual robot automatically updates its gait simulation model based on feedback from the physical robot. The physical robot, in turn, executes more precise movements based on the virtual test data generated by the updated gait simulation model. This process ensures high-precision, real-time, and robust gait testing between the virtual and physical robots.

[0050] This embodiment achieves dynamic adjustment and continuous optimization of gait data by constructing a closed-loop interaction mechanism between the virtual robot and the physical robot, ensuring that the gait simulation model can highly restore the actual performance of the physical robot. Real-time analysis and calibration of gait errors improve the stability and accuracy of gait, and significantly enhance the adaptability and motion performance of the physical robot in complex scenarios. The introduction of machine learning algorithms and control optimization technology makes the gait simulation model more flexible and robust, and can quickly respond to scene changes. Through the closed-loop data transmission and optimization process, the testing cost and risk of the physical robot are effectively reduced, the R&D cycle is shortened, and an efficient and reliable technical path is provided for improving the motion performance of the physical robot.

[0051] It should be noted that the gait simulation test process includes several preset time periods, each of which has a corresponding sub-gait scheme. Several sub-gait schemes form a complete gait scheme. As the gait simulation test progresses, the gait simulation model is updated based on the sub-gait schemes generated in each preset time period, ultimately achieving gait synchronization between the virtual robot and the physical robot.

[0052] In the preferred implementation of this embodiment, there are many methods for constructing an actual test environment, which are not limited here. For example, based on the gait test task, a virtual test environment is constructed; a gait test platform is constructed; wherein the gait test platform is used to indicate a platform capable of regulating the test parameters of the test environment; the test parameters include at least: test speed, and / or ground friction coefficient, and / or test slope, and / or right-angle turn; based on the virtual test environment, the gait test platform is controlled in a coordinated manner to generate an actual test environment corresponding to the virtual test environment.

[0053] Construct a standardized gait test platform; simulate the gait performance of the humanoid robot at different walking speeds by adjusting the platform speed of the gait test platform; simulate the gait adaptability of the humanoid robot in uphill and downhill environments by adjusting the platform inclination angle of the test platform; evaluate the stability of the humanoid robot under different friction conditions by adjusting the surface friction coefficient of the test platform; design a right-angle turning area of ​​the test platform, collect the accuracy and stability of the humanoid robot's turning path through sensors, and evaluate its ability to complete turning movements.

[0054] like Figure 4 1 and 2 are a top view and a side view of a test platform provided by an embodiment of the present invention.

[0055] S1, design a test platform with adjustable speed.

[0056] Design and implement a speed-adjustable gait test platform capable of simulating the gait performance of a humanoid robot at various speeds. The test platform's speed adjustment range is 0.1m / s to 3m / s, controlled by a motor drive system and equipped with a high-precision encoder for real-time speed feedback. This design enables evaluation of a humanoid robot's gait stability and speed adaptability at various speeds.

[0057] S2, the test platform needs to be equipped with a degree of freedom for adjusting the tilt angle.

[0058] The gait test platform incorporates an adjustable tilt angle, supporting a slope adjustment range of 0° to 30°. Slope adjustment is accomplished via an electro-hydraulic system, simulating walking and climbing performance on surfaces of varying slopes. This design allows for testing the gait adaptability and stability of humanoid robots under varying slopes, and for evaluating their locomotion performance in complex environments.

[0059] S3, set up a test table with adjustable friction coefficient.

[0060] By coating the platform surface with a layer of magnetorheological fluid to form a magnetorheological fluid-coated conveyor belt, combined with a magnetic field control device underneath, the surface friction coefficient can be dynamically adjusted. The friction coefficient can be precisely adjusted within a range of 0.1 to 1.0 to simulate different surface types (such as ice, snow, sand, mud, etc.). This dynamic friction surface design provides highly controllable experimental conditions for gait testing, helping to evaluate the robot's gait performance in various friction environments.

[0061] S4, the design includes a right-angle turn area.

[0062] A right-angle turning area was designed into the test platform to simulate the gait of a humanoid robot during cornering. Angle and position sensors were installed to monitor the robot's path accuracy and stability during cornering in real time. This design aims to test how the robot adjusts its gait during cornering to maintain stability and precision.

[0063] S5, design supporting control software to enable the test platform to automatically adjust parameters such as speed, slope, friction coefficient, etc. according to the settings of the virtual test environment.

[0064] The software seamlessly integrates with the virtual test environment, automatically configuring the test platform's physical parameters based on the test task's requirements. Furthermore, the accompanying software displays virtual or real-world test data in real time, helping users monitor the performance of the test platform during interaction with the humanoid robot and optimize the test process.

[0065] Therefore, by adding supporting control software, the test platform and the virtual test environment are deeply integrated. The platform's speed, slope, and friction coefficient can be automatically adjusted according to test requirements, reducing manual intervention and improving test efficiency and accuracy. The supporting control software's real-time feedback and adjustment capabilities optimize the test process, ensuring that each test task is performed under the optimal configuration. Furthermore, the supporting control software standardizes test procedures, improving the test platform's repeatability and adaptability, and providing comprehensive and scientific support for humanoid robot gait testing.

[0066] In a preferred implementation of this embodiment, the current preset time period includes several acquisition time points; the current virtual test data includes several groups of virtual test data; the current real test data includes several groups of real test data; each of the acquisition time points corresponds to a group of virtual test data and a group of real test data; the virtual test data includes several first parameters; the real test data includes several second parameters; for any acquisition time point within the current preset time period: based on the first parameter and the second parameter corresponding to the acquisition time point, determine the parameter deviation corresponding to the acquisition time point; based on the parameter deviation corresponding to each of the acquisition time points, determine the average value of the parameter deviation corresponding to the current preset time period; based on the parameter deviation corresponding to each of the acquisition time points, the average value of the parameter deviation, and the number of acquisition time points within the current preset time period, generate the variance of the parameter deviation; determine the variance of the parameter deviation as an indicator for evaluating the gait performance of the virtual robot.

[0067] For example, the first parameters include virtual gait key point positions, virtual joint angles, virtual zero-torque points, and virtual center-of-mass trajectory, etc.; the second parameters include real gait key point positions, real joint angles, real zero-torque points, and real center-of-mass trajectory, etc.

[0068] (1) Current gait key point position deviation

[0069] By recording the coordinate positions of the physical robot and the virtual robot at key gait points (such as starting, mid-section, and landing), their deviations are calculated, and the variance is used to evaluate the error stability.

[0070] Let P pyt (t) and P sim (t) represents the coordinates of the gait key points (such as the center of the sole and the hip joint) of the physical robot and the virtual robot at time t, respectively. The position deviation of any gait key point E p (t) is calculated by the following formula:

[0071] E p (t)=||P pyt (t)-P sim (t)|| Formula (1);

[0072] Calculate the variance of the current gait key point position deviation

[0073]

[0074] Where N represents the number of collection time points in the current preset time period; is the average value of the position deviation of the current gait key points, as follows:

[0075]

[0076] (2) Current joint angle deviation

[0077] set up and Represents the joint angles of the j-th joint (such as the hip joint) of the physical robot and the virtual robot at time t: Calculate the joint angle deviation between the physical robot and the virtual robot

[0078]

[0079] The variance of the current joint angle deviation calculate:

[0080]

[0081] if The smaller it is, the higher the gait accuracy of the joint.

[0082] (3) Current zero torque point deviation

[0083] The zero moment point reflects the stability of the robot during gait.

[0084] Let Z phy (t) and Z sim (t) represents the zero moment point coordinates of the physical robot and the virtual robot at time t, and the zero moment point deviation E ZMP (t) is shown in the following formula (6):

[0085] E ZMP (t)=||Z phy (t)-Z sim (t)|| Equation (6);

[0086] Calculate the variance of the current zero moment point deviation

[0087]

[0088] if If it is too large, it means that the zero moment point fluctuates greatly and the gait stability is poor.

[0089] (4) Current center of mass trajectory deviation (CoM)

[0090] Set CoM phy (t) and CoM sim (t) represents the coordinates of the center of mass trajectory of the physical robot and the virtual robot at time t, and the center of mass trajectory deviation E CoM (t) is shown in the following formula (8):

[0091] E CoM (t)=||CoM phy (t)-CoM sim (t)|| Equation (8);

[0092] Calculate the variance of the current center of mass trajectory deviation

[0093]

[0094] like If it is too large, it means that the posture of the humanoid robot is unstable when walking.

[0095] In a preferred implementation manner of this embodiment, based on the current virtual test data and the current real test data, an update operation is performed on the current gait simulation model to generate an updated gait simulation model; including: based on the current virtual test data and the current real test data, determining the current gait deviation corresponding to the virtual robot; wherein the current gait deviation at least includes: the current gait joint point position deviation, the current joint angle deviation, the current zero torque point deviation, and the current center of mass trajectory deviation; based on the current gait deviation and the current real test data, the gait error of the virtual robot is corrected to generate the starting gait of the virtual robot in the next preset time period; based on the current gait deviation and the current real test data, The optimization objective function of the current gait simulation model is determined based on the previous gait joint point position deviation, the current joint angle deviation, the current zero-torque point deviation, and the current center of mass trajectory deviation; according to the initial gait, the joint angles of the virtual robot are optimized and controlled based on PID gain tuning to generate optimized joint angles; based on the optimization objective function and the optimized joint angles, the constrained energy consumption corresponding to the reward function is determined; the gait scheme corresponding to the minimum constrained energy consumption is determined as the sub-gait scheme corresponding to the virtual robot in the next preset time period; based on the sub-gait scheme corresponding to the next time period, the current gait simulation model is updated to generate an updated gait simulation model.

[0096] For example: 1. Gait error correction

[0097] (1) For the parameter deviation corresponding to any acquisition time point, in order to minimize the parameter deviation between the physical gait and the virtual gait, an optimization objective function J is constructed;

[0098] J=α1||E p || 2 +α2||E θ || 2 +α3||E ZMP || 2 +α4||E CoM || 2

[0099] Formula (10);

[0100] Among them, α1, α2, α3, and α4 are weight coefficients used to balance the impact of different errors.

[0101] (2) Error correction

[0102] Joint angle correction: The error of the joint angle of the virtual robot at any acquisition time point is corrected to obtain the corrected joint angle. The specific calculation formula is shown in the following formula (11);

[0103]

[0104] Among them, θ new Represents the joint angle of the virtual robot after correction; θ old Indicates the joint angle of the virtual robot before correction.

[0105] Zero moment point (ZMP) correction: The zero moment point deviation E corresponding to any acquisition time point ZMP , if the zero moment point deviation E ZMP If the zero moment point deviation is greater than the preset threshold, the center of mass trajectory corresponding to the acquisition time point is corrected to make it closer to the stable support area; if the zero moment point deviation is not less than the preset threshold, the center of mass trajectory corresponding to the acquisition time point is not corrected. The center of mass trajectory correction formula is as follows:

[0106] CoM new (t) = CoM old (t)+k ZMP E ZMP (t) Formula (12);

[0107] Among them, CoM new (t) represents the trajectory of the center of mass of the virtual robot after correction; CoM old (t) represents the trajectory of the center of mass of the virtual robot before correction.

[0108] Gait key point position correction: Dynamic Time Warping (DTW) is used to match physical gait key points with virtual gait key points. The swing position is adjusted for the corresponding gait key points at any acquisition time point:

[0109] P new (t) = P old (t)+k p E p (t); Formula (13);

[0110] Among them, P new (t) represents the coordinates of the key points of the virtual robot’s gait after correction; P old (t) represents the coordinates of the key points of the virtual robot's gait before correction, and kp is the correction coefficient;

[0111] 2. Gait control strategy optimization

[0112] (1) PID gain tuning

[0113] For the next preset time period: If the joint angle deviation is large, the PD controller parameters can be adjusted: increasing the P gain can improve the response speed and reduce the deviation; appropriately adjust the D gain to prevent overshoot.

[0114] The joint angle optimization control rate is shown in the following formula:

[0115]

[0116] (2) Gait optimization based on reinforcement learning

[0117] Use reinforcement learning (such as PPO or TD3) to optimize the gait strategy: To optimize the gait, design a reward function:

[0118] R=-J-β∑|τ j | Formula (15);

[0119] Among them, β is the weight, τ j is the joint torque determined based on the optimized joint angle, and R is the constrained energy consumption.

[0120] After the physical robot performs a gait task, the virtual gait model is optimized based on the feedback performance data (such as gait stability, movement completion, joint angles, etc.), making it more closely aligned with the real physical system and improving gait accuracy, stability, and execution. Based on the error data analysis mentioned above, gait parameters are adjusted through error compensation methods to make the gait simulation model more similar to the physical robot.

[0121] like Figure 2 FIG. 1 is a flow chart of determining an optimal gait solution for a virtual robot in a virtual test environment according to an embodiment of the present invention.

[0122] Determining the optimal gait scheme of the virtual robot for the virtual test environment includes at least the following steps:

[0123] S201, obtaining several complete gait schemes based on the complete gait scheme generated by each gait simulation test in several gait simulation tests; wherein each complete gait scheme includes quasi-virtual test data formed by at least two sets of virtual test data and quasi-real test data formed by at least two sets of real test data.

[0124] S202 , for any complete gait scheme among the several complete gait schemes: determining a relative closeness between the quasi-virtual test data and the quasi-real test data corresponding to the complete gait scheme.

[0125] S203 : Obtaining a plurality of relative closenesses based on the relative closenesses corresponding to each complete gait scheme.

[0126] S204 , selecting a complete gait solution with the greatest relative closeness from the plurality of relative closenesses as the optimal gait solution for the virtual robot to perform a gait simulation test on the virtual test environment.

[0127] Specifically, based on the quasi-virtual test data and quasi-real test data corresponding to the complete gait scheme, the quasi-gait deviation of the complete gait scheme is determined to generate a decision matrix; wherein the quasi-gait deviation includes at least the quasi-gait joint point position deviation, the quasi-joint angle deviation, the quasi-zero moment point deviation, the quasi-center of mass trajectory deviation, and the minimum quasi-energy consumption; all the quasi-gait deviations in the decision matrix are normalized to generate a standard matrix formed by standardized values; corresponding weights are applied to each of the standardized values ​​in the standard matrix to generate an update matrix; positive ideal solutions and negative ideal solutions are calculated according to the update matrix, and the distances between each of the update matrices and the positive ideal solution and the negative ideal solution are calculated to obtain corresponding first distances and second distances; the relative closeness of the complete gait scheme is determined based on the first distance and the second distance.

[0128] For example, each completed gait solution corresponds to a set of test data. Multiple sets of test data are collected and analyzed using the TOPSIS method to comprehensively evaluate the robot's gait stability, accuracy, and energy consumption. The optimal gait solution is selected from several complete gait solutions. Key indicators such as gait joint position deviation, joint angle deviation, zero-torque point deviation, center of mass trajectory deviation, and minimum energy consumption are quantitatively evaluated. Vector normalization is used to calculate the standardized value of each indicator to ensure fair comparison of data of different dimensions.

[0129] TOPSIS evaluation process

[0130] (1) A decision matrix is ​​formed based on key indicators such as quasi-gait joint position deviation, quasi-joint angle deviation, quasi-zero moment point deviation, quasi-center of mass trajectory deviation, and minimum quasi-energy consumption corresponding to the complete gait scheme.

[0131] (2) Vector normalization processing, converting data of different dimensions into comparable standardized values:

[0132]

[0133] Among them, r ij is the normalized value after normalization, x ij is the raw data of key indicators, and m is the number of complete gait schemes.

[0134] (3) Assign different weights to different key indicators to highlight their impact on gait assessment:

[0135] v ij =w j ·r ij : Formula (17);

[0136] Among them, w jis the weight of each key indicator, which can be determined through expert experience or entropy weight method.

[0137] (4) Based on the updated matrix, calculate the positive ideal solution A + and negative ideal solution A - :

[0138] A + ={maxv ij |j∈J +}

[0139] A - ={minv ij |j∈J -} Formula (18);

[0140] Among them, J + is a benefit index (the larger the better, such as stability), J - It is a cost indicator (the smaller the better, such as energy consumption).

[0141] (5) Calculate the distance from the updated matrix to the positive ideal solution and the negative ideal solution respectively;

[0142] Calculate the Euclidean distance between the update matrix corresponding to the complete gait solution and the positive ideal solution:

[0143] Calculate the Euclidean distance between the update matrix corresponding to the complete gait solution and the negative ideal solution:

[0144] (6) Calculate relative closeness

[0145] Calculate the relative closeness C of the complete gait plan i :

[0146]

[0147] Among them, C i The value range is 0-1, and the closer the value is to 1, the better the complete gait plan.

[0148] According to the TOPSIS calculation results, the gait scheme with the highest relative closeness is selected as the optimal gait scheme.

[0149] This embodiment is described in detail below with reference to specific application scenarios.

[0150] An interactive testing method for gait simulation of a humanoid robot is applied to a testing platform; the testing platform includes a physical robot and a virtual robot; the physical robot is communicatively connected with the virtual robot; and the method includes at least the following steps:

[0151] S1, using kinematics and dynamics modeling tools, constructing a virtual robot motion model according to the specific structure of the physical robot; configuring a control algorithm corresponding to the gait test task for the virtual robot motion model to obtain a gait simulation model.

[0152] S2, controlling the current gait simulation model corresponding to the virtual robot to perform a gait simulation test in a virtual test environment within a current preset time period to generate current virtual test data; and sending the current virtual test data to the physical robot.

[0153] S3, constructing a virtual test environment according to the gait test task; constructing a gait test platform; wherein the gait test platform is used to indicate a platform capable of regulating the test parameters of the test environment; the test parameters include at least: test speed, and / or ground friction coefficient, and / or test slope, and / or right-angle turn; based on the virtual test environment, the gait test platform is controlled in a coordinated manner to generate an actual test environment corresponding to the virtual test environment.

[0154] S4, controlling the physical robot to perform corresponding gait operations according to the current virtual test data in an actual test environment corresponding to the virtual test environment, generating current real test data; and sending the current real test data to the virtual robot.

[0155] S5. Based on the current virtual test data and the current real test data, determine the current gait deviation corresponding to the virtual robot; wherein, the current gait deviation includes at least: the current gait joint point position deviation, the current joint angle deviation, the current zero-torque point deviation, and the current center of mass trajectory deviation; based on the current gait deviation and the current real test data, perform gait error correction on the virtual robot to generate the starting gait of the virtual robot in the next preset time period; based on the current gait joint point position deviation, the current joint angle deviation, the current zero-torque point deviation, and the current center of mass trajectory deviation, determine the optimization objective function of the current gait simulation model.

[0156] S6. Based on the initial gait, the virtual robot's joint angles are optimized and controlled based on PID gain tuning to generate optimized joint angles. Based on the optimization objective function and the optimized joint angles, a constrained energy consumption corresponding to a reward function is determined. The gait scheme corresponding to the minimum constrained energy consumption is determined as a sub-gait scheme corresponding to the virtual robot in a next preset time period. The current gait simulation model is updated based on the sub-gait scheme corresponding to the next time period to generate an updated gait simulation model.

[0157] S7, using the updated gait simulation model as the next gait simulation model corresponding to the next preset time period adjacent to the current preset time period, continuing to perform gait simulation testing in the next preset time period, and ending the gait simulation model update operation until the gait simulation test process is completed, and generating a complete gait plan corresponding to the virtual test environment.

[0158] S8. Based on the complete gait scheme generated by each gait simulation test in the multiple gait simulation tests, several complete gait schemes are obtained; wherein each of the complete gait schemes includes quasi-virtual test data formed by at least two sets of virtual test data and quasi-real test data formed by at least two sets of real test data.

[0159] S9, for any of the several complete gait schemes: based on the quasi-virtual test data and quasi-real test data corresponding to the complete gait scheme, determine the quasi-gait deviation of the complete gait scheme and generate a decision matrix; wherein the quasi-gait deviation at least includes the quasi-gait joint point position deviation, the quasi-joint angle deviation, the quasi-zero moment point deviation, the quasi-center of mass trajectory deviation, and the minimum quasi-energy consumption; normalize all the quasi-gait deviations in the decision matrix to generate a standard matrix formed by standardized values; apply corresponding weights to each of the standardized values ​​in the standard matrix to generate an update matrix; calculate the positive ideal solution and the negative ideal solution according to the update matrix, and calculate the distance between each of the update matrices and the positive ideal solution and the negative ideal solution respectively, to obtain the corresponding first distance and second distance; determine the relative closeness of the complete gait scheme based on the first distance and the second distance.

[0160] S10, based on the relative closeness corresponding to each of the complete gait schemes, obtain a plurality of relative closenesses; and select the complete gait scheme with the largest relative closeness from the plurality of relative closenesses as the optimal gait scheme for the virtual robot to perform a gait simulation test on the virtual test environment.

[0161] like Figure 3 FIG. 1 is a schematic diagram of an interactive testing system provided by an embodiment of the present invention.

[0162] The interactive testing system 300 includes at least a virtual simulation module 301, a physical robot control module 302, a data acquisition module 303, a platform control module 304, a virtual-physical synchronization module 305, a data analysis module 306, a visualization interaction module 307, and a test task control module 308. These modules seamlessly collaborate to achieve synchronized testing of virtual and physical gaits. The integration process prioritizes compatibility and openness in interface design, ensuring efficient and accurate data transmission and supporting real-time data exchange between modules.

[0163] The visual interaction module receives user test requests and sends them to the test task control module. The test task control module constructs a virtual test environment and sends it to the virtual simulation module and the platform control module. The virtual simulation module constructs a gait simulation model for the virtual robot and performs gait simulation tests in the virtual test environment, generating gait motion data. The virtual simulation module sends this gait motion data to the virtual-reality synchronization module. The data acquisition module collects dynamic environment data during the gait simulation process and sends it to the virtual-reality synchronization module. The virtual-reality synchronization module sends both the dynamic environment data and the gait motion data as virtual test data to the physical robot control module and the data analysis module, respectively.

[0164] The platform control module constructs an actual test environment corresponding to the virtual test environment; the physical robot control module controls the physical robot to perform corresponding gait operations in the actual test environment according to the virtual test data, generates gait motion data, and sends the gait motion data to the virtual-reality synchronization module; the data acquisition module collects dynamic environment data during the physical robot's gait operation and sends the dynamic environment data to the virtual-reality synchronization module; the virtual-reality synchronization module sends the gait motion data and dynamic environment data as real test data to the virtual simulation module and the data analysis module;

[0165] The data analysis module performs data analysis on the virtual test data and the real test data, and sends the data analysis results, the virtual test data, and the real test data to the visualization interaction module respectively.

[0166] The Visual Interaction Module provides a user-friendly interface, enabling users to monitor the gait synchronization between the virtual and physical robots in real time. This interface supports real-time display of test data, dynamic adjustment of test scenario parameters, and task flow configuration. It offers efficient test environment configuration capabilities, allowing users to quickly adapt to different testing needs.

[0167] The interactive visualization module also offers multi-dimensional data visualization capabilities, including dynamic trend charts, 3D simulation views, and heat maps, enabling users to analyze gait test results from various perspectives. The system supports on-demand analysis report export and utilizes data mining techniques to provide optimization recommendations, providing a scientific basis for improving gait performance.

[0168] The interactive test system in this embodiment is designed with a simple and user-friendly control system to ensure software stability and smooth operation. The modular design reduces the learning curve of the system and provides users with comprehensive operation instructions, online help documents and technical support, thereby improving the system's ease of use and user satisfaction.

[0169] This embodiment significantly improves test efficiency and accuracy by integrating virtual simulation, physical robot control, data acquisition, and platform control modules into a complete system. The intuitive interactive interface and real-time data feedback function enhance the user's controllability of the test process and reduce operational complexity. Multi-dimensional data visualization tools facilitate in-depth analysis of gait performance and provide a scientific basis for optimizing control algorithms, hardware design, and gait planning. The optimized user experience and comprehensive support functions lower the threshold for using the test system, improve the adaptability and application value of the test system in different scenarios, and provide reliable guarantees for the standardization and efficiency of robot gait testing.

[0170] The specific process is as follows.

[0171] Step 1: Virtual robot gait simulation.

[0172] Scenario configuration: Use virtual simulation software (such as Gazebo and V-REP) to create virtual test scenarios with various environmental conditions. These environments include various ground types (such as smooth ground, rough ground, sand, grass, etc.), different slope ranges (such as ramps from 0° to 30°), and simulated weather changes (such as rain and snow). The various elements of these virtual test scenarios can be flexibly adjusted through parameterized configuration, so that the construction of test scenarios can not only meet diverse needs but also ensure high test accuracy. These environmental settings can help study the robot's motion performance under various extreme conditions and provide data support for subsequent optimization.

[0173] Gait model construction: Based on the specific structure of the robot (for example, a bipedal robot or a quadruped robot), use kinematic and dynamic modeling tools (such as MATLAB / Simulink, Gazebo plug-in, etc.) to model and simulate the robot's joints, skeleton, and motion trajectory. This process not only needs to consider the robot's morphological structure, but also needs to be combined with the actual gait design requirements to establish the robot's motion model, simulating key dynamic parameters such as joint angles, stride length, and gait cycle. This method can ensure the naturalness and accuracy of the virtual robot's movement, and simulate the gait performance in the real environment as much as possible, especially under the influence of different loads and obstacles.

[0174] Control Algorithm Design: Design and implement control algorithms for the virtual robot's gait, including classic PID control and modern gait planning methods based on model predictive control (MPC). Gait planning algorithms generate various gaits, such as walking, running, and turning, based on the robot's task requirements. To ensure smooth movement of the virtual robot in complex environments, gait generation can be optimized based on time series algorithms or machine learning models, ensuring a natural and smooth trajectory and avoiding unnecessary vibration or instability.

[0175] Virtual Sensor Setup: Multiple sensors, such as ground contact force sensors, acceleration sensors, and velocity sensors, are integrated into the virtual environment to monitor the robot's dynamic behavior in real time. The data collected by these sensors can reflect physical indicators such as the robot's gait stability, gait patterns, ground contact conditions, and acceleration, providing important feedback for subsequent gait optimization and improvement. Furthermore, by simulating weather or ground changes in real time, the accuracy and response speed of the sensor setup can also simulate performance in different environments.

[0176] Real-time data transmission: Sensor data (e.g., gait cycle, acceleration, contact force, etc.) collected in the virtual environment is transmitted to the physical robot in real time via wireless communication protocols (e.g., Wi-Fi, Bluetooth, etc.) or real-time data buses (e.g., middleware systems like ROS). This synchronization of data transmission is crucial because it ensures data consistency between the virtual and physical robots during gait testing, facilitates analysis of discrepancies between the virtual model and physical execution, and provides a basis for optimization.

[0177] Step 2: Physical robot action feedback.

[0178] Physical Robot Selection: The physical robot should be a model with sufficient sensors and actuators, and possess at least 12 degrees of freedom (DOF). This ensures that its joint control system has high-precision control capabilities and can accurately execute gait data generated in the virtual environment. Furthermore, the physical robot must be equipped with a range of sensors, including joint sensors, accelerometers, and force sensors, to comprehensively monitor the robot's motion state and provide feedback.

[0179] Action Execution and Data Collection: In this step, the physical robot executes actions based on gait data generated in the virtual environment. During execution, the robot utilizes joint sensors, accelerometers, force sensors, and other devices to record motion data in real time. Key data includes gait stability, movement completion, ground contact force, and acceleration. This data can be used to evaluate the robot's gait performance in the physical environment and identify gaps and deficiencies between the virtual model and the real-world implementation.

[0180] Data Feedback and Analysis: The physical robot transmits real-time motion data to the virtual environment via high-speed communication protocols (such as Ethernet, CAN, and UDP). The virtual environment then uses data analysis tools to analyze this data in detail, assessing whether the physical robot exhibits gait errors during actual operation. This data may include inconsistent gait, joint angle deviations, and gait cycle instability. Based on this feedback, the virtual robot's gait model is further optimized.

[0181] Step 3: Synchronize the gaits of the virtual and physical robots.

[0182] Data Analysis: Gait data from both the physical and virtual robots is analyzed using data analysis tools (such as MATLAB, Pandas in Python, and the Numpy library). This analysis includes gait errors, gait inconsistencies, and joint angle deviations, with a focus on key metrics such as gait stability and movement completion for both the virtual and physical robots under different environmental conditions. This data analysis not only helps identify issues but also provides data support for the design of subsequent optimization algorithms.

[0183] Application of Optimization Algorithms: Optimization algorithms (such as genetic algorithms and particle swarm optimization) are used to adjust the virtual gait model. Optimization parameters include joint angles, stride length, and gait period. The goal is to ensure that the virtual robot's gait matches the physical robot's gait as closely as possible during execution. This optimization process may involve multiple cycles. After each optimization, the virtual model's data is retransmitted to the physical robot to continuously adjust the gait and gradually reduce the error between the virtual and physical robots.

[0184] Closed-loop interaction and continuous optimization: Optimized virtual gait data is then transmitted back to the physical robot, ensuring it continues to perform actions based on the optimized data. This process forms a closed-loop interaction mechanism between the virtual and physical robots. Through repeated optimization and verification, the gait gap between the virtual model and the physical robot will gradually narrow, thereby improving the accuracy and practicality of gait testing.

[0185] Step 4: Construction of standardized gait test platform, such as Figure 4 shown.

[0186] Step 5: Standardized gait test tasks and scenario construction.

[0187] Straight-line walking test: The robot walks in a straight line at a constant speed on a flat surface. During the test, the friction surface is adjusted to simulate different surface conditions (e.g., smooth floor, rough cement, wet surface, etc.). Gait stability indicators (e.g., step deviation, joint angle error), gait cycle, speed fluctuation, etc. are recorded to evaluate the robot's walking ability and stability under different surface conditions.

[0188] Endurance Testing: The robot is set to walk at various speeds for extended periods on an adjustable speed platform to test its battery life. The platform periodically changes slope and friction surface to simulate real-world energy consumption. During this process, the robot's travel time and distance are recorded to evaluate its battery performance and motion efficiency.

[0189] Variable Speed ​​Walking Test: The robot accelerates from a standstill to accelerated walking using a platform, setting various acceleration curves (constant acceleration, step acceleration, etc.), as well as deceleration from high-speed walking to a stop. This test captures gait changes, stability, and the impact of inertia on gait during acceleration and deceleration, assessing its dynamic gait adjustment capabilities.

[0190] Cornering Walk Test: This test assesses the robot's gait performance and adaptability during cornering, using a designed route that includes right-angle turns. The platform simulates cornering conditions with varying slopes and friction, requiring the robot to complete the turn at a specific speed. The test comprehensively evaluates the robot's performance on complex paths by recording trajectory deviation, center of gravity shift, and gait stability. This test helps optimize the robot's cornering control strategy, ensuring a smooth and stable gait at varying turn radii and speeds, and avoiding instability or imbalance during cornering.

[0191] Slope walking test: The slope walking test evaluates the robot's gait stability and climbing ability under slope changes by setting different inclination angles (0° to 30°). During uphill, downhill, and lateral walking tasks on slopes, the robot needs to automatically adjust its pace to cope with the changes in center of gravity and gait adjustments caused by different slopes. The test will monitor the robot's pace adjustment amplitude, climbing ability, stability, and energy consumption performance to help evaluate the robot's adaptability and energy efficiency under different slope conditions. With this data, the robot's gait algorithm can be further optimized to ensure its excellent performance in complex terrain.

[0192] Virtual Test Environment Construction: The goal of constructing a virtual test scenario is to create an environment that closely mirrors the actual test task, enabling simulation, validation, and optimization of the robot's gait performance. By using efficient virtual modeling tools such as Gazebo or Unity, virtual test scenarios can be designed with diverse floor materials, slopes, right-angle curves, and other features. These virtual test scenarios accurately simulate various environmental conditions likely to be encountered in the real world, providing a repeatable and efficient testing platform for validating the robot's gait control algorithms. Furthermore, the actual test platform can adjust and simulate different sections of the virtual test scenario in real time, ensuring consistency and synchronization between virtual and physical testing. The construction of virtual test scenarios not only enables rapid adjustment of environmental parameters (such as floor friction coefficient, slope, and obstacle distribution), but also allows for multiple experiments under different conditions to comprehensively evaluate the robot's gait performance and adaptability in diverse environments. The results of virtual testing provide crucial data support for actual physical testing, helping to further optimize the robot's gait control algorithm and improve its stability and reliability in complex and dynamic environments.

[0193] Step 6: Data collection and analysis.

[0194] Real-time data acquisition: Various sensors (such as ground contact force sensors, acceleration sensors, and velocity sensors) are used to record various data from both the virtual and physical robots during gait testing. This data includes key parameters such as gait stability, ground contact force, robot acceleration, velocity change, and gait cycle. The data acquisition system monitors the robot's behavior in real time under different environmental conditions and provides detailed motion feedback. By combining data from different sensors, a precise basis for gait adjustments can be provided, helping to identify potential stability issues, dynamic problems, or perception errors.

[0195] Data Analysis and Performance Evaluation: During the data collection process, data analysis methods are used to conduct in-depth analysis of the collected data. Common analysis methods include regression analysis, principal component analysis (PCA), and Fourier transforms to evaluate the robot's gait performance under different test conditions. Analysis focuses on gait stability, gait accuracy, adaptability and robustness, and energy consumption analysis.

[0196] Data visualization and report generation: After completing data collection and analysis, use visualization tools (such as MATLAB, Matplotlib in Python, Seaborn, and Plotly) to generate detailed charts and reports. These charts can include data trend charts for gait stability, speed changes, gait cycle, joint angle changes, and other aspects. When generating reports, it is important to highlight key analysis points, such as gait error analysis, dynamic adjustment capability, and energy consumption assessment. This will help decision makers quickly understand the test results and provide guidance for subsequent optimization.

[0197] This embodiment achieves dynamic adjustment and continuous optimization of gait data by constructing a closed-loop interaction mechanism between virtual and physical robots, ensuring that the virtual model can highly restore the actual performance of the physical robot. Real-time analysis and calibration of gait errors improve the stability and accuracy of gait, and significantly enhance the adaptability and motion performance of physical robots in complex scenarios. The introduction of machine learning algorithms and control optimization technologies makes the gait model more flexible and robust, and can quickly respond to scene changes. Through the closed-loop data transmission and optimization process, the testing cost and risk are effectively reduced, the R&D cycle is shortened, and an efficient and reliable technical path is provided for improving the robot's motion performance.

[0198] like Figure 5 FIG. 1 is a schematic structural diagram of an interactive testing device for gait simulation of a humanoid robot provided by an embodiment of the present invention.

[0199] An interactive test device for gait simulation of a humanoid robot, the device 500 is applied to a test platform; the test platform includes a physical robot and a virtual robot; the physical robot is in communication with the virtual robot; the device includes: a gait simulation test module 501 for controlling a current gait simulation model corresponding to the virtual robot to perform a gait simulation test in a virtual test environment within a current preset time period, generating current virtual test data; and sending the current virtual test data to the physical robot; a gait synchronization operation module 502 for controlling the physical robot to perform corresponding gait simulation tests according to the current virtual test data in an actual test environment corresponding to the virtual test environment. Gait operation, generating current real test data; and sending the current real test data to the virtual robot; a model update operation module 503, used to perform an update operation on the current gait simulation model based on the current virtual test data and the current real test data, to generate an updated gait simulation model; a generation module 504, used to use the updated gait simulation model as the next gait simulation model corresponding to the next preset time period adjacent to the current preset time period, and continue to perform gait simulation testing in the next preset time period, until the gait simulation test process is completed and the update operation of the gait simulation model is ended, and a complete gait plan corresponding to the virtual test environment is generated.

[0200] In a preferred implementation of this embodiment, the device also includes: a first construction module, used to construct a virtual test environment according to the gait test task; a second construction module, used to construct a gait test platform; wherein the gait test platform is used to indicate a platform that can regulate the test parameters of the test environment; the test parameters include at least: test speed, and / or ground friction coefficient, and / or test slope, and / or right-angle turn; a first generation module, used to control the gait test platform based on the virtual test environment, and generate an actual test environment corresponding to the virtual test environment.

[0201] In a preferred implementation of this embodiment, the device also includes: a third construction module, used to use kinematic and dynamic modeling tools to construct a virtual robot motion model according to the specific structure of the physical robot; a configuration module, used to configure a control algorithm corresponding to the gait test task for the virtual robot motion model to obtain a gait simulation model.

[0202] In a preferred implementation of this embodiment, the device also includes: a first acquisition module, used to obtain several complete gait schemes based on the complete gait scheme generated by each gait simulation test in several gait simulation tests; wherein each of the complete gait schemes includes quasi-virtual test data formed by at least two sets of virtual test data and quasi-real test data formed by at least two sets of real test data; a first determination module, used to determine, for any one of the several complete gait schemes: based on the quasi-virtual test data and quasi-real test data corresponding to the complete gait scheme, the relative closeness between the quasi-virtual test data and the quasi-real test data; a second acquisition module, used to obtain several relative closenesses based on the relative closeness corresponding to each of the complete gait schemes; a selection module, used to select the complete gait scheme with the largest relative closeness from the several relative closenesses as the optimal gait scheme for the virtual robot to perform gait simulation test on the virtual test environment.

[0203] In a preferred implementation of this embodiment, the first determination module includes: a first determination unit, used to determine the quasi-gait deviation of the complete gait scheme based on the quasi-virtual test data and quasi-real test data corresponding to the complete gait scheme, and generate a decision matrix; wherein the quasi-gait deviation at least includes quasi-gait joint point position deviation, quasi-joint angle deviation, quasi-zero moment point deviation, quasi-center of mass trajectory deviation, and minimum quasi-energy consumption; a normalization unit, used to normalize all quasi-gait deviations in the decision matrix to generate a standard matrix formed by standardized values; a generation unit, used to apply corresponding weights to each of the standardized values ​​in the standard matrix to generate an update matrix; a calculation unit, used to calculate the positive ideal solution and the negative ideal solution according to the update matrix, and calculate the distance between each of the update matrices and the positive ideal solution and the negative ideal solution respectively, to obtain the corresponding first distance and second distance; a second determination unit, used to determine the relative closeness of the complete gait scheme based on the first distance and the second distance.

[0204] In a preferred implementation manner of this embodiment, the model update operation module includes: a determination unit, used to determine the current gait deviation corresponding to the virtual robot based on the current virtual test data and the current real test data; wherein, the current gait deviation includes at least: the current gait joint point position deviation, the current joint angle deviation, the current zero torque point deviation, and the current center of mass trajectory deviation; a gait error correction unit, used to perform gait error correction on the virtual robot based on the current gait deviation and the current real test data, and generate the starting gait of the virtual robot in the next preset time period; a prediction unit, used to predict the sub-gait scheme corresponding to the virtual robot in the next preset time period based on the current gait deviation and the starting gait; a model update unit, used to perform an update operation on the current gait simulation model based on the sub-gait scheme corresponding to the next time period, and generate an updated gait simulation model.

[0205] In a preferred implementation of this embodiment, the prediction unit includes: a first determination subunit, used to determine the optimization objective function of the current gait simulation model based on the current gait joint point position deviation, the current joint angle deviation, the current zero torque point deviation, and the current center of mass trajectory deviation; an optimization control unit, used to optimize the joint angle of the virtual robot based on PID gain tuning according to the starting gait, and generate an optimized joint angle; a second determination subunit, used to determine the constraint energy consumption corresponding to the reward function based on the optimization objective function and the optimized joint angle; a third determination subunit, used to determine the gait scheme corresponding to the minimum constraint energy consumption as the sub-gait scheme corresponding to the virtual robot in the next preset time period.

[0206] In a preferred implementation of this embodiment, the current preset time period includes several acquisition time points; the current virtual test data includes several groups of virtual test data; the current real test data includes several groups of real test data; each of the acquisition time points corresponds to a group of virtual test data and a group of real test data; the virtual test data includes several first parameters; the real test data includes several second parameters; the device also includes: a second determination module, which is used to determine, for any acquisition time point within the current preset time period: based on the first parameter and second parameter corresponding to the acquisition time point, the parameter deviation corresponding to the acquisition time point; a third determination module, which is used to determine the average value of the parameter deviation corresponding to the current preset time period based on the parameter deviation corresponding to each of the acquisition time points; a second generation module, which is used to generate the variance of the parameter deviation based on the parameter deviation corresponding to each of the acquisition time points, the average value of the parameter deviation, and the number of acquisition time points within the current preset time period; and a fourth determination module, which is used to determine the variance of the parameter deviation as an indicator for evaluating the gait performance of the virtual robot.

[0207] The above-described device can execute the interactive testing method for humanoid robot gait simulation provided by one embodiment of the present invention, and possesses the functional modules and beneficial effects corresponding to executing the interactive testing method for humanoid robot gait simulation. For technical details not fully described in this embodiment, please refer to the interactive testing method for humanoid robot gait simulation provided by one embodiment of the present invention.

[0208] The present invention also provides an electronic device, comprising: a processor; a memory for storing executable instructions of the processor; the processor is used to read the executable instructions from the memory and execute the instructions to implement the interactive testing method for humanoid robot gait simulation described in the present invention.

[0209] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.

[0210] The computer program product may be written in any combination of one or more programming languages ​​to implement the program code for performing the operations of the embodiments of the present application, including object-oriented programming languages ​​such as Java, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0211] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, enable the processor to execute the steps of the method according to the following embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0212] The computer-readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can, for example, include but is not limited to a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0213] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this application are merely illustrative and not restrictive, and it should not be assumed that these advantages, strengths, and effects are required of each embodiment of this application. In addition, the specific details disclosed above are merely illustrative and facilitating understanding, and are not restrictive. The above details do not limit this application to necessarily being implemented using the above specific details.

[0214] The block diagrams of the devices, devices, equipment, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0215] It should also be noted that in the apparatus, device, and method of the present application, each component or each step can be decomposed and / or recombined, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0216] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0217] The above description has been provided for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

[0218] In the description of this specification, reference to the terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless otherwise inconsistent.

[0219] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0220] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. An interactive testing method for gait simulation of a humanoid robot, characterized in that: Applied to a test platform; the test platform includes a physical robot and a virtual robot; the physical robot is communicatively connected with the virtual robot; Controlling the current gait simulation model corresponding to the virtual robot to perform a gait simulation test in a virtual test environment within a current preset time period to generate current virtual test data; and sending the current virtual test data to the physical robot; Controlling the physical robot to perform corresponding gait operations according to the current virtual test data in an actual test environment corresponding to the virtual test environment to generate current real test data; and sending the current real test data to the virtual robot; performing an updating operation on the current gait simulation model based on the current virtual test data and the current real test data to generate an updated gait simulation model; The updated gait simulation model is used as the next gait simulation model corresponding to the next preset time period adjacent to the current preset time period, and the gait simulation test is continued in the next preset time period. The update operation of the gait simulation model is terminated only after the gait simulation test process is completed, and a complete gait plan corresponding to the virtual test environment is generated.

2. The method according to claim 1, characterized in that Also includes: Construct a virtual test environment based on the gait test task; Constructing a gait test platform; wherein the gait test platform is used to indicate a platform capable of regulating test parameters of a test environment; the test parameters include at least: test speed, and / or ground friction coefficient, and / or test slope, and / or right-angle turn; Based on the virtual test environment, the gait test platform is controlled in a coordinated manner to generate an actual test environment corresponding to the virtual test environment.

3. The method according to claim 1, characterized in that Also includes: Using kinematics and dynamics modeling tools, a virtual robot motion model is constructed according to the specific structure of the physical robot; A control algorithm corresponding to the gait test task is configured for the virtual robot motion model to obtain a gait simulation model.

4. The method according to claim 1, wherein Also includes: Based on the complete gait scheme generated in each of the multiple gait simulation tests, a plurality of complete gait schemes are obtained; wherein each of the complete gait schemes includes quasi-virtual test data formed by at least two sets of virtual test data and quasi-real test data formed by at least two sets of real test data; For any one of the several complete gait schemes: determining a relative closeness between the quasi-virtual test data and the quasi-real test data corresponding to the complete gait scheme; Based on the relative closeness corresponding to each of the complete gait schemes, a plurality of relative closenesses are obtained; The complete gait scheme with the largest relative closeness is selected from the plurality of relative closenesses as the optimal gait scheme for the virtual robot to perform a gait simulation test on the virtual test environment.

5. The method according to claim 4, characterized in that The step of determining the relative closeness between the quasi-virtual test data and the quasi-real test data based on the quasi-virtual test data and the quasi-real test data corresponding to the complete gait scheme comprises: Based on the quasi-virtual test data and the quasi-real test data corresponding to the complete gait scheme, determining the quasi-gait deviation of the complete gait scheme and generating a decision matrix; wherein the quasi-gait deviation at least includes quasi-gait joint point position deviation, quasi-joint angle deviation, quasi-zero moment point deviation, quasi-center of mass trajectory deviation, and minimum quasi-energy consumption; Normalizing all quasi-gait deviations in the decision matrix to generate a standard matrix formed by standardized values; Applying a corresponding weight to each of the standardized values ​​in the standard matrix to generate an update matrix; Calculating a positive ideal solution and a negative ideal solution according to the update matrix, and calculating the distance between each of the update matrices and the positive ideal solution and the negative ideal solution, respectively, to obtain corresponding first distances and second distances; A relative closeness of the complete gait plan is determined based on the first distance and the second distance.

6. The method according to claim 1, characterized in that The updating operation is performed on the current gait simulation model based on the current virtual test data and the current real test data to generate an updated gait simulation model; comprising: Determine a current gait deviation corresponding to the virtual robot based on the current virtual test data and the current real test data; wherein the current gait deviation includes at least: a current gait joint point position deviation, a current joint angle deviation, a current zero-torque point deviation, and a current center of mass trajectory deviation; Based on the current gait deviation and the current real test data, the gait error of the virtual robot is corrected to generate a starting gait of the virtual robot in the next preset time period; Based on the current gait deviation and the initial gait, predicting a sub-gait solution corresponding to the virtual robot in the next preset time period; An updating operation is performed on the current gait simulation model based on the sub-gait scheme corresponding to the next time period to generate an updated gait simulation model.

7. The method according to claim 6, characterized in that The method of predicting a sub-gait scheme corresponding to the virtual robot in the next preset time period based on the current gait deviation and the initial gait comprises: Determining an optimization objective function of the current gait simulation model based on the current gait joint point position deviation, the current joint angle deviation, the current zero moment point deviation, and the current center of mass trajectory deviation; According to the initial gait, optimizing and controlling the joint angles of the virtual robot based on PID gain tuning to generate optimized joint angles; Determining a constrained energy consumption corresponding to a reward function based on the optimization objective function and the optimized joint angle; The gait scheme corresponding to the minimum constraint energy consumption is determined as the sub-gait scheme corresponding to the virtual robot in the next preset time period.

8. The method according to claim 1, characterized in that The current preset time period includes a plurality of acquisition time points; the current virtual test data includes a plurality of groups of virtual test data; the current real test data includes a plurality of groups of real test data; each acquisition time point corresponds to a group of virtual test data and a group of real test data; the virtual test data includes a plurality of first parameters; and the real test data includes a plurality of second parameters; For any acquisition time point within the current preset time period: determining a parameter deviation corresponding to the acquisition time point based on the first parameter and the second parameter corresponding to the acquisition time point; Determining an average value of the parameter deviation corresponding to the current preset time period based on the parameter deviation corresponding to each of the acquisition time points; Generate a variance of the parameter deviation based on the parameter deviation corresponding to each acquisition time point, the average value of the parameter deviation, and the number of acquisition time points in the current preset time period; The variance of the parameter deviation is determined as an indicator for evaluating the gait performance of the virtual robot.

9. An interactive testing device for gait simulation of a humanoid robot, characterized in that: Applied to a test platform; the test platform includes a physical robot and a virtual robot; the physical robot is communicatively connected with the virtual robot; a gait simulation test module, configured to control a current gait simulation model corresponding to the virtual robot to perform a gait simulation test in a virtual test environment within a current preset time period, generate current virtual test data, and send the current virtual test data to the physical robot; a gait synchronization operation module, configured to control the physical robot to perform corresponding gait operations according to the current virtual test data in an actual test environment corresponding to the virtual test environment, generate current real test data, and send the current real test data to the virtual robot; a model updating operation module, configured to perform an updating operation on the current gait simulation model based on the current virtual test data and the current real test data, to generate an updated gait simulation model; A generation module is used to use the updated gait simulation model as the next gait simulation model corresponding to the next preset time period adjacent to the current preset time period, continue to perform gait simulation testing in the next preset time period, and end the updating operation of the gait simulation model until the gait simulation test process is completed, thereby generating a complete gait plan corresponding to the virtual test environment.

10. A computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images