Security test method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a security test method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring environment information and attribute information of a target object; performing behavior simulation on the environment information and the attribute information according to a motion analysis model, and determining behavior control data; executing the target behavior based on the behavior control data; obtaining interaction response data under the target behavior, and feeding back the interaction response data to the target simulation platform; and the target simulation platform is used for analyzing the security test result according to the interaction response data. By adopting the method, the accuracy of a safety test result can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle safety, and particularly to a safety testing method, device, computer device, computer-readable storage medium, and computer program product. Background Art

[0002] With the development of the ADAS (Advanced Driver Assistance System) and autonomous driving technologies, in order to ensure the safety and reliability of autonomous driving, it is necessary to test and evaluate intelligent functions such as pedestrian recognition, avoidance, and emergency braking in autonomous driving.

[0003] In the traditional technology, a pedestrian substitute with a fixed structure is used to simulate a pedestrian crossing the road, and it is jointly used with an autonomous driving vehicle for the test of autonomous driving. The pedestrian substitute is manually manipulated to move and collide with the autonomous driving vehicle, and the collision data of different parts of the pedestrian substitute at the time of collision is obtained through the collision between the pedestrian substitute and the autonomous driving vehicle. Then, based on the collision data, the autonomous driving vehicle is subjected to a safety test to obtain a safety test result.

[0004] However, in the traditional technology, due to the activity limitations of the pedestrian substitute, the accuracy of simulating pedestrian behavior by the pedestrian substitute in the safety test is poor, and further the accuracy of the safety test for the autonomous driving vehicle is poor. Summary of the Invention

[0005] Based on this, it is necessary to provide a safety testing method, device, computer device, computer-readable storage medium, and computer program product for the above technical problems.

[0006] In a first aspect, the present application provides a safety testing method, including:

[0007] Obtain environmental information and attribute information of a target object;

[0008] Perform behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data;

[0009] Execute a target behavior based on the behavior control data;

[0010] Obtain interaction response data under the target behavior, and feed back the interaction response data to a target simulation platform; the target simulation platform is used to analyze a safety test result according to the interaction response data.

[0011] In one embodiment, the performing behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data includes:

[0012] Perform behavior prediction on the environmental information and the attribute information according to the motion analysis model to obtain the behavior data and target posture at each time step in a preset time series before collision;

[0013] Based on the inverse kinematics algorithm, the behavior data and the target posture at each time step, calculate the control of each component of the target object to obtain the behavior control data of each component before collision.

[0014] In one embodiment, the motion analysis model includes a first inference structure and a second inference structure;

[0015] The performing behavior prediction on the environmental information and the attribute information according to the motion analysis model to obtain the behavior data and target posture at each time step in a preset time series before collision includes:

[0016] Determine the first behavior probability of the target object according to the first inference structure and the environmental information;

[0017] Based on the second inference structure, the first behavior probability and the attribute information, determine the second behavior probability, and based on the second behavior probability, determine the behavior data and target posture at the current time step;

[0018] If the environmental information is continuously updated, execute the step of determining the first behavior probability of the target object according to the first inference structure and the environmental information until the environmental information does not change, to obtain the behavior data and the target posture at each time step in a preset time series before collision.

[0019] In one embodiment, the components of the target object include a mobile platform and a plurality of joint components; the behavior control data includes the motion speed and motion direction of the mobile platform at each time step, and the component control data of each joint component;

[0020] The calculating the control of each component of the target object based on the inverse kinematics algorithm, the behavior data and the target posture at each time step to obtain the behavior control data of each component before collision includes:

[0021] Based on the behavior data at each time step, determine the motion speed and motion direction of the mobile platform at each time step;

[0022] Based on the inverse kinematics algorithm and the target posture at each time step, determine the corresponding component control data of each joint component.

[0023] In one embodiment, determining the component control data corresponding to each of the joint components based on the inverse kinematics algorithm and the target pose at each time step includes:

[0024] For each time step, obtain the first coordinate information of each joint component of the target pose at the previous time step and the second coordinate information of each joint component of the target pose at the current time step;

[0025] Based on the inverse kinematics algorithm and the distance between the first coordinate information and the second coordinate information, determine the component control data corresponding to each joint component at each time step.

[0026] In one embodiment, the target behavior includes a motion behavior and a pose behavior; performing the target behavior based on the behavior control data includes:

[0027] Perform the motion behavior based on the motion speed, the motion direction, and the mobile platform, and perform the pose behavior based on the component control data and each joint component.

[0028] In a second aspect, the present application also provides a safety testing device, including:

[0029] An acquisition module, configured to acquire environmental information and attribute information of a target object;

[0030] A simulation module, configured to perform behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data;

[0031] An execution module, configured to perform a target behavior based on the behavior control data;

[0032] A feedback module, configured to acquire interaction response data under the target behavior and feed back the interaction response data to a target simulation platform; the target simulation platform is configured to analyze a safety test result according to the interaction response data.

[0033] In one embodiment, the simulation module is specifically configured to perform behavior prediction on the environmental information and the attribute information according to a motion analysis model to obtain behavior data and a target pose at each time step in a preset time series before a collision;

[0034] Calculate the control of each component of the target object based on the inverse kinematics algorithm, the behavior data, and the target pose at each time step to obtain the behavior control data of each component before the collision.

[0035] In one embodiment, the motion analysis model includes a first inference structure and a second inference structure; the simulation module is specifically configured to determine the first behavior probability of the target object according to the first inference structure and the environmental information;

[0036] Based on the second inference structure, the first behavior probability, and the attribute information, determine the second behavior probability, and based on the second behavior probability, determine the behavior data and the target pose at the current time step;

[0037] If the environmental information is continuously updated, execute the step of determining the first behavior probability of the target object according to the first inference structure and the environmental information until the environmental information does not change, and obtain the behavior data and the target pose at each time step in the preset time series before the collision.

[0038] In one embodiment, the components of the target object include a mobile platform and a plurality of joint components; the behavior control data includes the movement speed and movement direction of the mobile platform at each time step, and the component control data of each joint component; the simulation module is specifically configured to determine the movement speed and movement direction of the mobile platform at each time step based on the behavior data at each time step;

[0039] Based on the inverse kinematics algorithm and the target pose at each time step, determine the component control data corresponding to each joint component.

[0040] In one embodiment, the simulation module is specifically configured to, for each time step, obtain the first coordinate information of each joint component of the target pose at the previous time step and the second coordinate information of each joint component of the target pose at the current time step;

[0041] Based on the inverse kinematics algorithm and the distance between the first coordinate information and the second coordinate information, determine the component control data corresponding to each joint component at each time step.

[0042] In one embodiment, the target behavior includes a motion behavior and a pose behavior; the execution module is specifically configured to execute the motion behavior based on the movement speed, the movement direction, and the mobile platform, and execute the pose behavior based on the component control data and each joint component.

[0043] In a third aspect, the present application further provides a safety testing system, the system includes:

[0044] A target object, which is used to obtain environmental information and attribute information of the target object; perform behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data; execute a target behavior based on the behavior control data; obtain interaction response data under the target behavior, and feedback the interaction response data to a target simulation platform;

[0045] The target simulation platform is used to analyze security test results according to the interaction response data.

[0046] In one embodiment, the target object is further used to send the environmental information and the attribute information to the target simulation platform, receive the behavior control data fed back by the target simulation platform, and execute a target behavior based on the behavior control data; obtain interaction response data under the target behavior, and feedback it to the target simulation platform;

[0047] The target simulation platform is further used to receive the environmental information and the attribute information sent by the target object, perform behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine the behavior control data, and send the behavior control data to the target object; receive the interaction response data fed back by the target object, and analyze security test results according to the interaction response data.

[0048] In a fourth aspect, the present application further provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0049] Obtain environmental information and attribute information of a target object;

[0050] Perform behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data;

[0051] Execute a target behavior based on the behavior control data;

[0052] Obtain interaction response data under the target behavior, and feedback the interaction response data to a target simulation platform; the target simulation platform is used to analyze security test results according to the interaction response data.

[0053] In a fifth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0054] Obtain environmental information and attribute information of a target object;

[0055] Perform behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data;

[0056] Execute a target behavior based on the behavior control data;

[0057] Obtain interaction response data under the target behavior, and feedback the interaction response data to a target simulation platform; the target simulation platform is used to analyze security test results based on the interaction response data.

[0058] In a sixth aspect, the present application also provides a computer program product, including a computer program, which when executed by a processor implements the following steps:

[0059] Obtain environmental information and attribute information of a target object;

[0060] Perform behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data;

[0061] Execute a target behavior based on the behavior control data;

[0062] Obtain interaction response data under the target behavior, and feedback the interaction response data to a target simulation platform; the target simulation platform is used to analyze security test results based on the interaction response data.

[0063] The above security test method, device, computer device, computer-readable storage medium, and computer program product obtain environmental information and attribute information of a target object; perform analysis and processing on the environmental information and the attribute information according to a motion analysis model to determine behavior control data of the target object before a collision; execute a target posture based on the control data; obtain interaction response data of the target object in the target posture, and feedback the interaction response data to a target simulation platform; the target simulation platform is used to analyze security test results based on the interaction response data. By using this method, behavior control data is obtained through a motion analysis model, and the target object executes a target posture through the behavior control data, optimizing the dynamic response ability of the target object's behavior, making the behavior of the target object in security tests more in line with the real scenario, improving the motion complexity and authenticity that the target object can achieve, and obtaining interaction response data of the target object in a more realistic target posture, which can improve the accuracy of collision response data, thereby improving the accuracy of security test results. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for describing the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0065] Figure 1 is an application environment diagram of the security testing method in an embodiment;

[0066] Figure 2 is a schematic flowchart of the security testing method in an embodiment;

[0067] Figure 3 is a schematic flowchart of determining behavior control data based on a motion analysis model and an inverse kinematics algorithm in an embodiment;

[0068] Figure 4 is a schematic flowchart of determining behavior data and target postures at each time step in an embodiment;

[0069] Figure 5 is a schematic flowchart of looping the first - layer Bayesian inference and the second - layer Bayesian inference in an embodiment;

[0070] Figure 6 is a schematic flowchart of determining component control data at each time step based on an inverse kinematics algorithm in an embodiment;

[0071] Figure 7 is a schematic flowchart of the specific process of determining component control data in an embodiment;

[0072] Figure 8 is a schematic architecture diagram of a security testing system in an embodiment;

[0073] Figure 9 is a structural block diagram of a security testing device in an embodiment;

[0074] Figure 10 is an internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0075] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0076] The security testing method provided by the embodiments of the present application can be applied to, for example, Figure 1A security testing system in the application environment shown. Among them, the target object 102 communicates with the target simulation platform 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or can be placed on the cloud or other network servers. The target object 102 obtains environmental information and attribute information of the target object; the target object 102 performs behavior simulation on the environmental information and attribute information according to the motion analysis model to determine behavior control data; the target object 102 executes the target behavior based on the behavior control data; after the target object 102 is collided, it obtains the interaction response data under the target behavior and feeds the interaction response data back to the target simulation platform 104; the target simulation platform is used to analyze the security test results according to the interaction response data. Among them, the target object 102 can be a pedestrian substitute, and the pedestrian substitute can be composed of a pedestrian dummy, an omnidirectional mobile platform, and a control module. Among them, the pedestrian dummy is composed of a humanoid skeleton, a flexible outer cortex, a multi-degree-of-freedom joint module, sensors and a control unit, simulating the main joints of the human body (such as shoulders, elbows, waists, knees, ankles, etc.), and each joint has no less than 2-3 degrees of freedom, has the degrees of freedom of the human joints and can represent the evasive postures according to the walking trajectory or action mode set by the target simulation platform 104 in the physical environment; the omnidirectional motion platform is equipped with Mecanum wheels or other omnidirectional wheels, providing 360° omnidirectional motion ability, and supporting the target object 102 to perform multi-directional motion; the control module includes the joint control and motion control of the target object 102, and realizes the precise control of the joints of the pedestrian dummy through servo motors and encoders. The target simulation platform 104 can be set on an independent physical server, or can be set on a server cluster or distributed system composed of multiple physical servers, or can also be set on a cloud server providing cloud computing services.

[0077] In an exemplary embodiment, as Figure 2 shown, a security testing method is provided. Taking the target object in Figure 1 as an example, the following steps 202 to step 208 are included. Among them:

[0078] Step 202, obtain environmental information and attribute information of the target object.

[0079] In the embodiment of the present application, a main control board is integrated in the target object. The main control board can be a high-performance microcontroller or an embedded processor, responsible for the acquisition and processing of sensor data and controlling the actions of servo motors. The communication interface of the target object supports a variety of communication protocols to exchange data and transmit instructions with the omnidirectional mobile platform and the upper-layer target simulation platform.

[0080] Integrate a sensing and encoding unit (millimeter-wave radar, lidar, ultrasonic, or visual marking system) on the target object to obtain environmental information through sensors. Alternatively, the test scenario also includes off-site sensors, and the target object can interact with the off-site sensors to obtain environmental information. The attribute information of the target object can be pre-loaded into the target object, or multiple preset attribute information can be set. This attribute information includes the age, gender, personality settings, etc. corresponding to the simulated pedestrian, so as to simulate different types of pedestrians and conduct safety tests, enriching the types of objects for safety tests.

[0081] The environmental information includes the first position information of the object under test, speed information, the second position information of the target object, attribute information, integrated posture, orientation, and environmental illumination and road condition information. Taking the object under test as a test vehicle as an example, the target object obtains the first position information of the test vehicle and the second position information of the target object in real time through its own integrated sensors or off-site sensors, and calculates the relative distance according to the first position information and the second position information and relative speed . The target object inputs kinematic quantities such as the relative distance and relative speed as well as the integrated posture, orientation, and environmental illumination and road condition information of the target object into the encoder for encoding to obtain environmental information and attribute information in a multi-dimensional encoded form. Optionally, the target object can also integrate the vehicle acceleration and jerk characteristics of the test vehicle to form environmental information and attribute information in a multi-dimensional encoded form as the input of the motion analysis model.

[0082] Step 204: Perform behavior simulation on the environmental information and attribute information according to the motion analysis model to determine the behavior control data.

[0083] Among them, the behavior control data includes the motion direction, motion speed, motion acceleration of the target object, and the angle changes of each joint in the target object.

[0084] In the embodiments of the present application, taking the offline deployment of the motion analysis model as an example for illustration, through the main control board built into the target object, the motion analysis model can be deployed to the target object, and the target object can achieve offline safety testing. Specifically, the written control program is burned into the control board of the target object through the programming interface by the local program for offline execution. The control board of the target object independently executes the pre-loaded control program without relying on an external network connection, realizing completely autonomous behavior execution. The motion trajectory and posture of the target object can be pre-learned during the training process of the motion analysis model. Taking the hip joint control of the target object as an example for illustration, the motion state data of the real pedestrian and vehicle interaction under the complete time axis is used as sample data. The sample data includes step frequency (the walking rhythm of the real pedestrian, for example, the number of steps per minute), walking speed (linear speed and angular speed), upper and lower limb postures (the posture changes of the upper limb arm swing and the lower limb leg movement during walking, including joint angles and positions), and attribute information such as the age, gender, and personality setting of the target object.

[0085] To meet different test requirements, the motion analysis model can also be deployed in real time, and it can be flexibly switched between the offline deployment and real-time deployment modes. When switching modes, the main control board of the target object will automatically load the corresponding control program and communication protocol. When switching to the real-time mode, the main control board will automatically connect to the 5G network and establish a data transmission channel with the cloud target simulation platform to realize the dynamic adjustment of the active avoidance action; when switching to the offline mode, the main control board will disconnect the network connection and instead execute the pre-loaded control program. Through the flexible switching mechanism of offline deployment and real-time deployment, the present invention can adapt to different test requirements and ensure the behavior simulation and test effects of the target object in various environments. The offline mode provides an independent and stable test environment, while the real-time mode realizes more accurate and dynamic behavior adjustment through the computing power of the cloud platform. The two complement each other and jointly improve the test efficiency and accuracy of the autonomous driving system.

[0086] In an exemplary embodiment, for the training of the motion analysis model, first, the target object obtains a sample data set; the sample data set includes multiple sample data and the sample labels corresponding to each sample data, and each sample data is the behavior data and attribute information of the sample object. The target object inputs each sample data into the motion analysis model to be trained, and performs behavior prediction on each sample data according to the motion analysis model to obtain initial behavior control data; determines the loss value of the motion analysis model to be trained according to the initial behavior control data and the sample label; adjusts the parameters of the motion analysis model to be trained according to the loss value until the loss value meets the preset loss condition, and obtains the trained motion analysis model.

[0087] Based on the real-time given environmental information and attribute information, the target object uses a motion analysis model to simulate the street-crossing behavior of the target object according to the current environmental information and the attribute information of the target object, and obtains the behavior control data of the target object. For example, if the tested vehicle approaches, the target object may choose to avoid or accelerate across the street; if the tested vehicle moves away, the target object may choose to continue moving.

[0088] Step 206, execute the target behavior based on the behavior control data.

[0089] In the embodiment of the present application, the target object executes the joint angle instruction in the behavior control data through a servo motor and an encoder, and controls the step frequency and walking speed of the target object by adjusting the rotation speed of the servo motor, so that the target object executes the target behavior according to the behavior control data, and makes the motion state of the target object conform to the pedestrian behavior corresponding to the attribute information when executing the target behavior, so as to simulate the walking rhythm of a real pedestrian.

[0090] Step 208, obtain the interaction response data under the target behavior, and feedback the interaction response data to the target simulation platform.

[0091] Among them, the target simulation platform is used to analyze the safety test results according to the interaction response data.

[0092] In the embodiment of the present application, the interaction response data includes the response data collected by the sensors of the target object after the target object is collided by the tested vehicle when the target object conducts a passive test on the tested vehicle, and also includes the interaction response data of the tested vehicle after the target object executes the target behavior when the target object conducts an active safety test on the tested vehicle. For example, when the target object executes target behaviors such as "avoidance", the interaction response data collected by the sensors of the tested vehicle and / or the target object.

[0093] Taking the passive test as an example for illustration, the target object includes multiple sensors for detecting the interaction response data of each part of the target object. After the target object is collided, the target object obtains the interaction response data of each part when the target object executes the target behavior through the sensors. The interaction response data includes changes in the motion state, such as changes in the position, speed, and acceleration corresponding to each part. Then, the target object uploads the interaction response data collected after the collision to the target simulation platform, and the target simulation platform analyzes the safety test results of the target object and the tested vehicle according to the collision response data, including the damage degree of the target object, the distribution of collision energy, and the safety performance of the tested vehicle (such as collision energy absorption and occupant protection effect).

[0094] In the above safety testing method, behavior control data is obtained through a motion analysis model, and the target object is made to execute a target posture through the behavior control data, optimizing the dynamic response ability of the target object's behavior, making the behavior of the target object in the safety test more in line with the real scenario, improving the motion complexity and authenticity that the target object can achieve. The target object obtains interaction response data in a more realistic target posture, which can improve the accuracy of collision response data, and further improve the accuracy of the safety test results.

[0095] In an exemplary embodiment, as Figure 3 shown, step 204 includes steps 302 to 304. Among them:

[0096] Step 302, perform behavior prediction on the environmental information and attribute information according to the motion analysis model to obtain the behavior data and target posture at each time step in a preset time series before collision.

[0097] In the embodiment of the present application, when the target object uses the current attribute information as the simulation object according to the motion analysis model, the potential actions of the target object under different environmental conditions (for example, "crossing the street", "avoiding", "standing still") are evaluated through a behavior decision algorithm, and before the target object collides with the vehicle under test, the behavior data and target posture of the target object are predicted step by step in a preset time series. The behavior data includes motion parameters such as the speed, acceleration, and direction of the target object; the target posture includes the spatial position and orientation of each part of the target object (such as the head, torso, and limbs). Finally, the behavior data and target posture at each time step are generated for subsequent inverse kinematics calculation.

[0098] Step 304, calculate the control of each part of the target object based on the inverse kinematics algorithm, the behavior data and target posture at each time step to obtain the behavior control data of each part before collision.

[0099] In the embodiment of the present application, the target object uses the behavior data and target posture at each time step as the input data of the inverse kinematics algorithm. The inverse kinematics algorithm calculates the motion parameters of each part of the target object (such as joints and limbs) according to the target posture, and through the inverse kinematics equation, solves the angles and positions of each joint to achieve the target posture. Finally, the target object converts the inverse kinematics calculation result into the behavior control data of each part of the target object through the inverse kinematics algorithm. The behavior control data includes the drive signals of each joint, for example, the rotation speed and torque of the servo motor, etc., to generate the behavior control data of each part before collision, which is used to drive the motion mechanism of the target object.

[0100] In this embodiment, by combining environmental information and target object attribute information, high-precision behavior prediction is achieved through a motion analysis model, and the behavior control data of each component is calculated in real time based on the inverse kinematics algorithm, which can improve the action accuracy of the target object and thus improve the accuracy of safety testing.

[0101] In an exemplary embodiment, the motion analysis model includes a first inference structure and a second inference structure. As Figure 4 shown, step 302 includes steps 402 to 406. Among them:

[0102] Step 402, determine the first behavior probability of the target object according to the first inference structure and environmental information.

[0103] In the embodiment of the present application, the motion analysis model adopts a two-layer Bayesian inference structure. The first inference structure and the second inference structure are the first-layer Bayesian inference and the second-layer Bayesian inference respectively. The first-layer Bayesian inference is used to take the environmental information as the likelihood input to determine the first behavior probability of the target object's tendency to start an active behavior. Specifically, the target object inputs the preprocessed and encoded multi-dimensional encoded form of environmental information into the first inference structure. The first inference structure determines the probability distribution of the target object's possible active behavior according to the external manifestations such as the facing direction and moving trend of the target object (for example, when the vehicle under test approaches, the target object is more likely to produce an active behavior of "avoidance"), and further calculates and determines the likelihood function of the target object , as the first behavior probability. represents the probability distribution of the target object's selection of a certain behavior (such as crossing the street, avoidance, etc.) given the evidence (such as relative distance, relative speed, etc.).

[0104] Step 404, determine the second behavior probability based on the second inference structure, the first behavior probability and the attribute information, and determine the behavior data and target posture at the current time step based on the second behavior probability.

[0105] In the embodiment of the present application, the second inference structure serves as the subjective decision-making model (prior) of the pedestrian, describing the posture and avoidance behavior probabilities under different emergency scenarios and different risk / benefit trade-offs. Finally, through the combination of the likelihood function (the first behavior probability) output by the first inference structure and the prior, the posterior decision probability, that is, the second behavior probability, is obtained. The target object calculates the behavior data and target posture at the current time step according to the second behavior probability to control the movement of the target object and thus interact with the vehicle under test.

[0106] First, the second inference structure defines the prior probability according to the attribute information of the target object or different emergency scenarios , where is a decision parameter. For example, if the personality is set to be cautious and safety is more emphasized, a higher risk weight may be assigned. If the personality is set to be aggressive and efficiency is more pursued, a higher return weight may be assigned. In an emergency situation, it may be necessary to cross the street faster, thus increasing the return weight. In a non-emergency situation, safety may be more emphasized, increasing the risk weight. Furthermore, a utility function U is established, comprehensively considering the potential benefits (time saved, need to reach the destination) and risks (collision danger, injury probability) brought by crossing the street. According to different emergency levels or environmental variables, corresponding weights are given to the return and risk:

[0107] (1)

[0108] Where, represents the return weight, represents the time saved by crossing the street, represents the risk weight, represents the risk of crossing the street.

[0109] The target object describes the prior probability of the decision parameter through a preset probability distribution (for example, normal distribution or Beta distribution). of the prior probability . For example, taking the decision parameter as the probability that the target object chooses to cross the street, taking the Beta distribution as an example for illustration, the prior probability is:

[0110] (2)

[0111] Where, is the parameter adjusted according to the personality setting and the emergency scenario.

[0112] The target object combines the first behavior probability , the prior probability through the second inference structure, and uses Bayes' theorem to further determine a more accurate posterior probability (i.e., the second behavior probability):

[0113] (3)

[0114] The target object determines the specific behavior data and target posture of the target object at the current time step based on the second behavior probability and a preset threshold. For example, when the second behavior probability is greater than the preset threshold, the target object will choose to execute an avoidance action.

[0115] In an exemplary embodiment, when the vehicle under test approaches a target object (e.g., a pedestrian) at a certain speed, the target object needs to make an avoidance decision based on the current environmental information and its own attribute information. The target object uses a second inference structure to evaluate the risk / benefit of the current scenario. For example, whether the speed of the approaching vehicle poses a threat. By combining the prior knowledge of the second inference structure and the likelihood function output by the first inference structure through a double-layer Bayesian inference model, a posterior decision probability is generated. This posterior decision probability reflects the confidence of the target object in choosing an avoidance behavior, that is, the second behavior probability. For example, "the probability of lateral movement avoidance is 80%, and the probability of staying in place is 20%". At the same time, based on the posterior decision probability of the second inference structure, the target object determines the behavior data at the current time step. For example, "perform a lateral movement avoidance action", and the target posture corresponding to "perform a lateral movement avoidance action", that is, the behavior data and body posture such as the movement speed and direction during lateral movement avoidance, including the displacement and angle of each joint, so as to realize the coordinated control of the mobile platform and each joint component.

[0116] Step 406, if the environmental information is continuously updated, execute the step of determining the first behavior probability of the target object according to the first inference structure and the environmental information until the environmental information does not change, and obtain the behavior data and target posture at each time step in the preset time series before collision.

[0117] In the embodiment of the present application, as Figure 5 shown, during the movement of the target object, the target object continuously updates its decision according to the continuous approach or deceleration information of the vehicle (the first inference structure and the second inference structure are executed in a loop), forming a closed loop until the target object has a collision interaction with the vehicle under test, so as to ensure that the target object can dynamically respond to the vehicle during the test.

[0118] In this embodiment, the accurate simulation and prediction of the behavior of the target object are realized through the double-layer Bayesian inference composed of the first inference structure and the second inference structure. First, use the first inference structure to determine the first behavior probability of the target object according to the environmental information, capture its behavior tendency in a specific environment, and combine the second inference structure to take into account the attribute information of the target object and the risk-benefit trade-off in different situations. Through comprehensive evaluation by the utility function, a more accurate posterior decision probability is obtained, so as to determine the specific behavior and posture of the target object. While considering environmental factors, the individual characteristics and situational adaptability of the target object are also incorporated, improving the accuracy and authenticity of pedestrian behavior simulation of the target object and the accuracy of safety testing of the autonomous driving system.

[0119] In an exemplary embodiment, the components of the target object include a mobile platform and a plurality of joint components; the behavior control data includes the movement speed and movement direction of the mobile platform in each time step, and the component control data of each joint component; asFigure 6 As shown in Figure 6 , step 304 includes steps 602 to 604, where:

[0120] Step 602: Based on the behavior data at each time step, determine the movement speed and movement direction of the mobile platform at each time step.

[0121] In the embodiments of the present application, the target object determines the movement speed and movement direction of the mobile platform at each time step based on the behavior data at each time step. The behavior data includes the behavior selections of the target object at specific time steps, such as "crossing the street", "avoiding", "waiting", etc. Each behavior corresponds to a different movement pattern. Further, the target object calculates the movement speed and movement direction according to the movement pattern. For example, when the movement pattern is "crossing the street", the target object moves towards the opposite side of the road, and the speed may be relatively fast; when the movement pattern is "avoiding", the target object may move to the side with a moderate speed. Optionally, the target object has physical limitations of dynamic constraints. For example, the maximum speed, acceleration, etc. of the target object have preset maximum values to ensure that the generated movement speed and movement direction conform to the dynamic constraints of the human body.

[0122] Step 604: Based on the inverse kinematics algorithm and the target pose at each time step, determine the component control data corresponding to each joint component.

[0123] In the embodiments of the present application, when simulating the behavior of the target object, it is necessary to precisely control the movement of its each joint to achieve natural and realistic actions. The target object determines the component control data corresponding to each joint component based on the inverse kinematics algorithm and the target pose at each time step. The target pose refers to the limb pose that the target object expects to reach at each time step, including the positions and orientations of various parts of the body. By applying the inverse kinematics algorithm, the angles of each joint can be calculated according to the target pose of the target object. At the same time, considering the movement range and physical limitations of the joints, it is ensured that the solved joint angles are feasible. In order to make the actions more natural, the joint angles can be smoothed to reduce mutations and achieve smooth movement transitions. The component control data can be the joint angle commands of the servo motors of each joint component, and the joint angle commands are used to control the rotation speed of the servo motors.

[0124] In this embodiment, by designing an omnidirectional movement platform with high degrees of freedom and multiple joint structures, the target object can simulate more diverse and complex pedestrian movement behaviors, including side movement, backward movement, fast paces, etc. This multi-degree-of-freedom movement control ability greatly improves the richness and realism of the simulation scenario, and calculates the component control data of each joint component through the inverse kinematics algorithm, further improving the accuracy of the target object's simulation of the posture and movement of pedestrians and the accuracy of the safety test of the autonomous driving system.

[0125] In an exemplary embodiment, such asFigure 7 As shown, step 604 includes steps 702 to 704. Among them:

[0126] Step 702, for each time step, obtain the first coordinate information of each joint component of the target pose at the previous time step and the second coordinate information of each joint component of the target pose at the current time step.

[0127] In the embodiment of the present application, for each time step, the target object needs to obtain the first coordinate information of each joint component of the target pose at the previous time step and the second coordinate information of each joint component of the target pose at the current time step. The first coordinate information and the second coordinate information reflect the pose change of the target object between consecutive time steps and serve as the basis for calculating the control data of each joint component.

[0128] Step 704, based on the inverse kinematics algorithm and the distance between the first coordinate information and the second coordinate information, determine the component control data corresponding to each joint component at each time step.

[0129] In the embodiment of the present application, the target object determines the component control data corresponding to each joint component at each time step based on the inverse kinematics algorithm and the distance between the first coordinate information and the second coordinate information. First, the target object calculates the coordinate difference between the first coordinate information and the second coordinate information of each joint component, that is, the displacement vector of each joint component from the previous time step to the current time step, as the distance between the first coordinate information and the second coordinate information. Then, the target object solves the distance between the first coordinate information and the second coordinate information according to the inverse kinematics algorithm, calculates the angles or positions that each joint needs to adjust to achieve the transition from the previous pose to the current pose, and in order to make the movement more natural, the target object smooths the change of the joint angles, avoids sudden drastic changes, and ensures that the calculated joint angles conform to the movement range limits of the joints, and finally obtains the component control data corresponding to each joint component at each time step. Furthermore, the servo motors of each joint component control the motor speed according to the component control data to control the target object to perform the target behavior.

[0130] In this embodiment, by obtaining the coordinate information of the joint components at the previous time step and the current time step and using the inverse kinematics algorithm to calculate the joint control data, the natural motion simulation of the target object is realized. It not only considers the behavior decision of the target object, but also ensures the authenticity and effectiveness of the simulation through precise motion control, providing strong support for the safety test of the autonomous driving system.

[0131] In an exemplary embodiment, the target behavior includes motion behavior and pose behavior, and step 206 includes step 2061. Among them:

[0132] Step 2061: Perform a motion behavior based on the motion speed, motion direction, and the mobile platform, and perform an attitude behavior based on the component control data and each joint component.

[0133] In the embodiment of the present application, according to the behavior data of each time step, the motion speed and motion direction of the target object are calculated. The behavior data is used to control the direction and wheel rotation speed of the mobile platform, so that the target object moves at a predetermined speed and direction. At the same time, through the inverse kinematics algorithm, according to the coordinate information of the joint components in the previous time step and the current time step, the component control data of each joint component is calculated. This control data is used to accurately control the rotation speed of the servo motors in each joint of the target object, so that it acts according to the expected attitude behavior.

[0134] In this embodiment, the coordinated work of the mobile platform and the joint components ensures that the target object can naturally and realistically exhibit the predetermined behaviors and postures in the simulation environment, so as to achieve dynamic interaction with the autonomous driving vehicle, and then interact with the vehicle under test in the real behaviors and postures of the target object, improving the accuracy of the safety test.

[0135] In an exemplary embodiment, as Figure 1 shown, a safety test system is provided, and the system includes:

[0136] A target object 102, configured to obtain environmental information and attribute information of the target object; perform behavior simulation on the environmental information and attribute information according to the motion analysis model to determine behavior control data; perform a target behavior based on the behavior control data; obtain interaction response data under the target behavior, and feedback the interaction response data to the target simulation platform;

[0137] A target simulation platform 104, configured to analyze the safety test result according to the interaction response data.

[0138] In a specific embodiment, as Figure 8 shown, the safety test system includes an application layer, a platform layer, a data layer, and a hardware layer. The application layer includes a 3D visualization platform, a control dashboard, and test scenario definition. The 3D visualization platform is used to display the test environment in a 3D interface; the control dashboard provides a user interface for monitoring and controlling the entire test process, and real-time displays test data and system status; the test scenario definition allows users to create and configure various test scenarios, including traffic environments, pedestrian behaviors, etc., and supports the definition of virtual and real vehicle test scenarios.

[0139] The platform layer includes a virtual simulation test module, a real vehicle test / acquisition module, a safety risk assessment module, and a data post-processing module. The virtual simulation test module is used to conduct safety tests in a virtual environment, simulating various traffic situations and pedestrian behaviors; the real vehicle test / acquisition module is used to conduct tests on actual vehicles and collect data in the real environment; the safety risk assessment module is used to analyze the safety test results based on the interaction response data feedback by the target object and determine the safety risk assessment results; the data post-processing module is used to process and analyze the collected data to generate test reports, including data cleaning and annotation, etc.

[0140] The data layer includes a distributed database and a data storage system. The distributed database is used to store real vehicle test data, usage environment data, digital twin data, and safety simulation data. The data storage system includes data management and data security to build a secure and confidential system.

[0141] The hardware layer includes a target object (pedestrian substitute), a vehicle under test, road section detection, and communication network equipment. Among them, the vehicle under test includes in-vehicle sensors and an ADAS (Advanced Driver Assistance Systems) system, the road section detection includes video sensors and lidar, and the communication network equipment supports data transmission and communication between modules during the test process.

[0142] In an exemplary embodiment, the target object 102 is further configured to send environmental information and attribute information to the target simulation platform 104, receive the behavior control data fed back by the target simulation platform 104, and execute the target behavior based on the behavior control data; after the target object 102 is collided, obtain the interaction response data under the target behavior and feed it back to the target simulation platform 104;

[0143] The target simulation platform 104 is further configured to receive the environmental information and attribute information sent by the target object 102, perform behavior simulation on the environmental information and attribute information according to the motion analysis model, determine the behavior control data, and send the behavior control data to the target object 102; receive the interaction response data fed back by the target object 102 and analyze the safety test results according to the interaction response data.

[0144] In the embodiments of the present application, the motion analysis model can also be deployed to a target simulation platform, and security testing can be performed through real-time communication between the target simulation platform and the target object. Specifically, the target object interacts with the target simulation platform in the cloud through a 5G network to achieve dynamic behavior simulation of virtual-real integration. The simulated pedestrian model in the target simulation platform and the target object (physical model) are made to be real-time consistent in terms of posture, speed, and behavior through digital twin technology. Sensors of the target object (such as IMU (Inertial Measurement Unit), force sensors, and vision sensors) collect data in real time and upload it to the target simulation platform in the cloud through a 5G network. The target simulation platform in the cloud generates avoidance instructions for the target object, including control instructions such as acceleration, steering angle, avoidance posture, and step frequency adjustment, based on the received data, the motion state of the vehicle under test, sensor data, environmental information, and the operation results of the active safety system model of the vehicle under test, and sends them to the target object for execution through a 5G network, and the simulated pedestrian models of the physical and virtual systems maintain synchronous consistency in state. Cloud computing can also be replaced by edge computing devices, and the calculation and data processing are performed locally, improving the real-time performance and accuracy of data processing, enhancing the stability of the system, and reducing the requirement for network bandwidth.

[0145] The target simulation platform uses a 3D engine to achieve three-dimensional visualization of virtual scene elements, including the test road, the vehicle under test, a pedestrian dummy (target object), traffic lights, road signs, and surrounding infrastructure. The target simulation platform maps the physical data collected in the real world into an adjustable, repeatable, and scalable simulation test scenario in the virtual environment. There is two-way information transfer between the virtual environment and the physical environment. The perception data of the physical world (vehicle sensor data, pedestrian detection data, roadside perception system data) is input into the virtual model in real time.

[0146] At the same time, high-bandwidth and low-latency wireless communication enables real-time data interconnection between the virtual platform and physical test equipment. 5G or advanced communication protocols provide the basis for two-way high-speed data interaction between the target simulation platform and on-site equipment, enabling the physical system and the digital environment to be synchronized at the millisecond level. The control strategy in the virtual scene can be transmitted and synchronized back to the physical system through a 5G network to drive real devices to perform specific actions or verify the effectiveness of the strategy.

[0147] It should be understood that although the steps in the flowcharts involved in the above embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0148] Based on the same inventive concept, an embodiment of the present application also provides a security testing device for implementing the security testing method involved above. The implementation solution provided by this device to solve the problem is similar to the implementation solution described in the above method. Therefore, the specific limitations in one or more embodiments of the security testing device provided below can refer to the limitations on the security testing method in the above text, and will not be repeated here.

[0149] In an exemplary embodiment, as Figure 9 shown, a security testing device 900 is provided, including: an acquisition module 901, a simulation module 902, an execution module 903, and a feedback module 904, where:

[0150] The acquisition module 901 is used to acquire environmental information and attribute information of the target object;

[0151] The simulation module 902 is used to perform behavior simulation on the environmental information and attribute information according to the motion analysis model to determine behavior control data;

[0152] The execution module 903 is used to execute the target behavior based on the behavior control data;

[0153] The feedback module 904 is used to acquire interaction response data under the target behavior and feedback the interaction response data to the target simulation platform; the target simulation platform is used to analyze the security testing result according to the interaction response data.

[0154] In one of the embodiments, the simulation module 902 is specifically used to perform behavior prediction on the environmental information and attribute information according to the motion analysis model to obtain behavior data and target postures at each time step in a preset time series before collision;

[0155] Calculate the control of each component of the target object based on the inverse kinematics algorithm, the behavior data at each time step, and the target posture to obtain the behavior control data of each component before collision.

[0156] In one embodiment, the motion analysis model includes a first inference structure and a second inference structure; the simulation module 902 is specifically configured to determine the first behavior probability of the target object according to the first inference structure and the environmental information;

[0157] Based on the second inference structure, the first behavior probability, and the attribute information, determine the second behavior probability, and based on the second behavior probability, determine the behavior data and the target pose at the current time step;

[0158] If the environmental information is continuously updated, execute the step of determining the first behavior probability of the target object according to the first inference structure and the environmental information until the environmental information does not change, and obtain the behavior data and the target pose at each time step in the preset time series before the collision.

[0159] In one embodiment, the components of the target object include a mobile platform and multiple joint components; the behavior control data includes the motion speed and motion direction of the mobile platform at each time step, and the component control data of each joint component; the simulation module 902 is specifically configured to determine the motion speed and motion direction of the mobile platform at each time step based on the behavior data at each time step;

[0160] Based on the inverse kinematics algorithm and the target pose at each time step, determine the component control data corresponding to each joint component.

[0161] In one embodiment, the simulation module 902 is specifically configured to, for each time step, obtain the first coordinate information of each joint component of the target pose at the previous time step and the second coordinate information of each joint component of the target pose at the current time step;

[0162] Based on the inverse kinematics algorithm and the distance between the first coordinate information and the second coordinate information, determine the component control data corresponding to each joint component at each time step.

[0163] In one embodiment, the target behavior includes a motion behavior and a pose behavior; the execution module 903 is specifically configured to execute the motion behavior based on the motion speed, the motion direction, and the mobile platform, and execute the pose behavior based on the component control data and each joint component.

[0164] Each module in the above safety testing device can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0165] In an exemplary embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 10As shown in the figure. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a security test method. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the computer device housing, or an external keyboard, touchpad, or mouse, etc.

[0166] Those skilled in the art can understand that Figure 10 the structure shown in the figure is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0167] In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the following steps are implemented:

[0168] Obtain environmental information and attribute information of the target object;

[0169] Perform behavior simulation on the environmental information and attribute information according to the motion analysis model to determine behavior control data;

[0170] Execute the target behavior based on the behavior control data;

[0171] Obtain interaction response data under the target behavior and feedback the interaction response data to the target simulation platform; the target simulation platform is used to analyze the security test results according to the interaction response data.

[0172] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0173] Perform behavior prediction on the environmental information and attribute information according to the motion analysis model to obtain the behavior data and target posture at each time step in a preset time series before collision;

[0174] Calculate the control of each component of the target object based on the inverse kinematics algorithm, the behavior data and target posture at each time step to obtain the behavior control data of each component before collision.

[0175] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0176] Determine the first behavior probability of the target object according to the first inference structure and environmental information;

[0177] Determine the second behavior probability based on the second inference structure, the first behavior probability and attribute information, and determine the behavior data and target posture at the current time step based on the second behavior probability;

[0178] If the environmental information is continuously updated, execute the step of determining the first behavior probability of the target object according to the first inference structure and environmental information until the environmental information does not change, to obtain the behavior data and target posture at each time step in a preset time series before collision.

[0179] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0180] Based on the behavior data at each time step, determine the motion speed and motion direction of the mobile platform at each time step;

[0181] Determine the component control data corresponding to each joint component based on the inverse kinematics algorithm and the target posture at each time step.

[0182] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0183] For each time step, obtain the first coordinate information of each joint component of the target posture in the previous time step and the second coordinate information of each joint component of the target posture in the current time step;

[0184] Determine the component control data corresponding to each joint component in each time step based on the distance between the inverse kinematics algorithm, the first coordinate information and the second coordinate information.

[0185] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0186] Execute motion behaviors based on the motion speed, motion direction, and mobile platform, and execute pose behaviors based on component control data and each joint component.

[0187] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0188] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0189] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0190] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0191] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this application.

[0192] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A safety testing method, characterized in that: The method is applied to a target object, and the method comprises: Obtaining environmental information and target object attribute information; Performing behavior simulation on the environmental information and the attribute information according to the motion analysis model to determine behavior control data; executing a target behavior based on the behavior control data; The interactive response data under the target behavior is obtained, and the interactive response data is fed back to the target simulation platform; the target simulation platform is used to analyze the safety test results according to the interactive response data.

2. The method according to claim 1, characterized in that The performing behavior simulation on the environmental information and the attribute information according to the motion analysis model to determine the behavior control data includes: Performing behavior prediction on the environmental information and the attribute information according to the motion analysis model to obtain behavior data and target posture at each time step in a preset time series before collision; Based on the inverse kinematics algorithm, the behavior data of each time step and the target posture, the control of each component of the target object is calculated to obtain the behavior control data of each component before the collision.

3. The method according to claim 2, characterized in that The motion analysis model includes a first reasoning structure and a second reasoning structure; The step of performing behavior prediction on the environment information and the attribute information according to the motion analysis model to obtain behavior data and target posture at each time step in a preset time sequence before the collision includes: Determine a first behavior probability of the target object according to the first reasoning structure and the environmental information; Determine a second behavior probability based on the second reasoning structure, the first behavior probability and the attribute information, and determine the behavior data and target posture of the current time step based on the second behavior probability; If the environmental information is continuously updated, the step of determining the first behavior probability of the target object according to the first inference structure and the environmental information is performed until the environmental information does not change, thereby obtaining the behavior data and the target posture at each time step in a preset time series before the collision.

4. The method according to claim 2, characterized in that: The components of the target object include a mobile platform and a plurality of joint components; the behavior control data include the movement speed and movement direction of the mobile platform in each time step, and component control data of each joint component; The calculation of the control of each component of the target object based on the inverse kinematics algorithm, the behavior data of each time step and the target posture to obtain the behavior control data of each component before the collision includes: Determine the movement speed and movement direction of the mobile platform at each time step based on the behavior data at each time step; The component control data corresponding to each joint component is determined based on an inverse kinematics algorithm and the target posture at each time step.

5. The method according to claim 4, characterized in that The determining of the component control data corresponding to each joint component based on the inverse kinematics algorithm and the target posture of each time step includes: For each of the time steps, obtaining first coordinate information of each of the joint components of the target posture at the previous time step and second coordinate information of each of the joint components of the target posture at the current time step; Based on an inverse kinematics algorithm and the distance between the first coordinate information and the second coordinate information, component control data corresponding to each of the joint components in each of the time steps is determined.

6. The method according to claim 4, characterized in that The target behavior includes motion behavior and posture behavior; and executing the target behavior based on the behavior control data includes: The motion behavior is performed based on the motion speed, the motion direction and the mobile platform, and the posture behavior is performed based on the component control data and each of the joint components.

7. A safety testing system, characterized in that: The system comprises: A target object is used to obtain environmental information and attribute information of the target object; perform behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data; execute a target behavior based on the behavior control data; obtain interactive response data under the target behavior, and feed the interactive response data back to a target simulation platform; The target simulation platform is used to analyze the safety test results according to the interactive response data.

8. The system according to claim 7, characterized in that The target object is also used to send the environment information and the attribute information to the target simulation platform, receive the behavior control data fed back by the target simulation platform, and execute the target behavior based on the behavior control data; Acquire interactive response data under the target behavior and feed it back to the target simulation platform; The target simulation platform is also used to receive the environment information and the attribute information sent by the target object, perform behavior simulation on the environment information and the attribute information according to the motion analysis model, determine the behavior control data, and send the behavior control data to the target object; The interactive response data fed back by the target object is received, and a security test result is analyzed according to the interactive response data.

9. A safety testing device, characterized in that: The device comprises: An acquisition module is used to obtain environmental information and attribute information of target objects; A simulation module, used for performing behavior simulation on the environmental information and the attribute information according to a motion analysis model to determine behavior control data; An execution module, configured to execute a target behavior based on the behavior control data; A feedback module is used to obtain interactive response data under the target behavior and feed back the interactive response data to a target simulation platform; the target simulation platform is used to analyze the safety test results according to the interactive response data.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

Images