Driving safety field model construction method and device, control method and system

Abstract

This invention relates to a method, apparatus, control method, and system for constructing a driving safety field model. The model construction method includes the following steps: defining the dynamic field generated by a single obstacle; calculating the sum of the dynamic fields generated by all obstacles within a preset area around the vehicle as the vehicle's risk potential energy; establishing a mapping equation between the risk potential energy and the reciprocal of the collision time; based on historical driving trajectory data, iteratively searching and optimizing the parameters in the mapping equation with the optimization objective of maximizing the linear fit integration between the risk potential energy and the reciprocal of the collision time, obtaining the optimal parameter set, and determining the driving safety field model; and updating the risk threshold of the driving safety field model according to real-time environmental parameters. This invention establishes a linear mapping between risk potential energy and kinematic indicators, enabling risk assessment to have a clear physical standard, and uses data-driven methods to solve the problem of difficulty in quantitatively determining model parameters, thereby improving modeling accuracy.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation and autonomous driving cooperative control technology, specifically relating to a method, device, control method and system for constructing a driving safety field model. Background Technology

[0002] The Driving Safety Field (DSF) is an important theoretical model in the field of intelligent connected vehicles and autonomous driving in recent years. It regards each traffic participant on the road as a field source, and these field sources generate a continuously distributed risk value in the surrounding space. It solves the problem that traditional discrete index evaluation has a single dimension and is difficult to handle complex scenarios with multiple vehicle interactions.

[0003] However, existing driving safety field models have the following limitations: (1) The parameters in the existing driving safety field model usually rely on expert experience or manual debugging, and lack scientific automated calibration methods.

[0004] (2) The calculated potential energy value has an ambiguous physical meaning and does not directly establish a mathematical correspondence with the industry-recognized safety indicators.

[0005] (3) In the existing autonomous driving control layer, the risk threshold is mostly set using a fixed value strategy, which has poor environmental adaptability and is prone to frequent false triggering or missed triggering under complex working conditions.

[0006] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention discloses a method, apparatus, control method, and system for constructing a driving safety field model.

[0008] The technical solutions adopted in the embodiments of the present invention are as follows: A method for constructing a driving safety field model includes the following steps: Define the dynamic field generated by a single obstacle, and calculate the sum of the dynamic fields generated by all obstacles within a preset area around the vehicle, as the risk potential energy of the vehicle. Establish a mapping equation between the risk potential energy and the reciprocal of the collision time; Obtain historical driving trajectory data of your own vehicle or vehicles of the same model as your own vehicle; Based on the historical driving trajectory data, with the optimization objective of maximizing the linear fitting integration degree between risk potential energy and the reciprocal of collision time, the parameters in the mapping equation are iteratively searched and optimized to obtain the optimal parameter set. Substitute the optimal parameter set into the mapping equation to determine the driving safety field model; Collect real-time environmental parameters and update the risk threshold of the driving safety field model based on the real-time environmental parameters.

[0009] A further technical solution is that the dynamic field of the single obstacle is represented by equation (1): (1) in, Represents the dynamic field of a single obstacle; They are respectively the bicycle and the first Euclidean distance and relative velocity between obstacles; Indicates the first The field strength ratio coefficient of each obstacle; Indicates the first The angle between the heading angle of the obstacle and the line connecting the vehicle; The equivalent mass coefficient of the obstacle. For the car and the first The equivalent mass coefficient of each obstacle; For speed sensitivity index, The distance decay is a power of 1. This is the anisotropic correction function.

[0010] The further technical solution is that the risk potential energy is expressed as equation (3): (3) in, N represents the vehicle's potential energy at risk; N represents the number of obstacles. Calculated using equation (4): (4) in, Here is the directional anisotropy coefficient. Indicates the number of obstacles. Indicates the first The angle between the heading angle of the obstacle and the line connecting the vehicle and the obstacle. This serves as the baseline reference length.

[0011] A further technical solution is that the mapping equation is expressed as equation (5): (5) in, The slope of the mapping. This is a static risk bias; This is the countdown to the collision time.

[0012] A further technical solution is that the historical driving trajectory data includes the relative speed, relative position, and collision time between the vehicle and the obstacle.

[0013] The further technical solution is to calibrate the parameter set using the following process: collection Group of natural driving data samples Construct the least squares optimization objective function according to equation (6). : (6) in, For sample index, The total number of samples, For parameter smoothing penalty term; To control the penalty intensity coefficient; The parameter vector to be calibrated, including the unknown parameters to be calibrated. ; The particle swarm optimization algorithm is used to optimize the objective function. The parameters to be determined are iteratively optimized.

[0014] A further technical solution is that the real-time environmental parameter is the road surface adhesion coefficient fed back by the sensor, and the risk threshold is corrected in real time according to equation (9): (9) in, The road surface adhesion coefficient, As a risk threshold, This is the revised risk threshold. This is the risk threshold before correction. It is an environmentally sensitive factor.

[0015] A vehicle safety field model construction device, comprising: The risk potential energy construction module is configured to construct the dynamic field generated by a single obstacle and calculate the sum of the dynamic fields generated by all obstacles within a preset area around the vehicle as the risk potential energy of the vehicle. The linear mapping construction module is configured to establish the mapping equation between the risk potential energy and the reciprocal of the collision time; The data acquisition module is configured to acquire historical driving trajectory data of the vehicle itself or vehicles of the same model as the vehicle itself, as well as acquire real-time environmental parameters; The parameter co-optimization module is configured to, based on the historical driving trajectory data, take maximizing the linear fitting integration degree between risk potential energy and the reciprocal of collision time as the optimization objective, iteratively search and optimize the parameters in the mapping equation to obtain the optimal parameter set. The driving safety field calculation module is configured to substitute the optimal parameters into the mapping equation to determine the driving safety field model; The threshold evolution and risk cost map generation module is configured to update the risk threshold based on the real-time environmental parameters and generate a continuous risk cost map for the entire scenario based on the optimal parameter set.

[0016] An autonomous driving control method includes the following steps: Collect real-time environmental parameters, monitor the vehicle's status, and perceive the motion characteristics of the vehicle and dynamic obstacles within the surrounding preset area; Calculate the reciprocal of the collision time based on the motion characteristic information; The motion feature information and the calculated reciprocal of the collision time are input into the pre-built autonomous driving safety field model to calculate the risk potential energy of the vehicle. When the risk potential energy is greater than or equal to the preset risk threshold, automatic emergency braking or forward collision warning is triggered. The driving safety field model is constructed using the following method: Define instantaneous risk potential energy, establish a mapping equation between the instantaneous risk potential energy and the reciprocal of the collision time, and based on historical driving trajectory data, take maximizing the linear fitting integration degree between risk potential energy and the reciprocal of the collision time as the optimization objective, iteratively search and optimize the parameters in the mapping equation to obtain the optimal parameter set, and substitute the optimal parameter set into the mapping equation to determine the driving safety field model. Based on the real-time environmental parameters, update the risk threshold; based on the optimal parameter set, generate a continuous risk cost map for the entire scenario.

[0017] An autonomous driving control system includes: The environmental perception and status acquisition layer is configured to collect real-time environmental parameters, monitor the vehicle's status, and perceive the motion characteristics of the vehicle and dynamic obstacles within the surrounding preset area. The kinematic evaluation index calculation layer is configured to calculate the reciprocal of the collision time based on the motion characteristic information. The calibration layer is configured to define risk potential energy, establish a mapping equation between the risk potential energy and the reciprocal of the collision time, and based on historical driving trajectory data, with the optimization objective of maximizing the linear fitting integration degree between the risk potential energy and the reciprocal of the collision time, iteratively search and optimize the parameters in the mapping equation to obtain the optimal parameter set. The optimal parameter set is then substituted into the mapping equation to determine the driving safety field model. The threshold evolution and risk cost map generation layer is configured to update the risk threshold based on the real-time environmental parameters and generate a continuous risk cost map for the entire scenario based on the optimal parameter set output by the calibration layer. The decision planning and active safety application layer is configured to input the calculated reciprocal of the collision time into the autonomous driving safety field model, output the risk potential energy of the vehicle, and determine whether the risk potential energy is greater than or equal to the risk threshold. If so, it triggers automatic emergency braking or forward collision warning; otherwise, it performs path planning based on the risk cost map and searches for a path with potential energy depression.

[0018] An electronic terminal includes a processor and a memory, the memory storing a computer program, the processor invoking the computer program to execute either the steps of the construction method or the steps of the control method.

[0019] A readable storage medium storing a computer program that, when invoked by a processor, executes either the steps of the construction method or the steps of the control method.

[0020] A computer program product includes a computer program / instructions that, when executed by a processor, implement the steps of the construction method or the control method.

[0021] The beneficial effects of the embodiments of the present invention are as follows: (i) The driving safety field model construction method and system proposed in this invention defines a dynamic field and risk potential energy based on kinematic characteristics, and then establishes a linear mapping relationship between risk potential energy and the reciprocal of collision time, giving the abstract potential energy a clear physical meaning. The potential energy gradient can accurately reflect the collision urgency at the physical level, enabling risk assessment to have a clear physical standard. It uses data-driven methods to solve the problem of model parameters being difficult to quantitatively determine, and provides a more accurate and robust theoretical basis for autonomous driving.

[0022] (II) The autonomous driving control method and system proposed in this invention are based on the constructed driving safety field model. Since the driving safety field model has been linearly calibrated by physical benchmarks and a dynamic risk threshold evolution mechanism based on the calibration slope has been designed, it can adaptively adjust the warning and control boundaries according to the road surface adhesion coefficient. Through dynamic threshold compensation, the driving safety and comfort of autonomous vehicles under complex weather conditions are effectively improved, and the trust between people and vehicles during autonomous driving is further enhanced. Attached Figure Description

[0023] Figure 1 This is a flowchart of the method for constructing a driving safety field model proposed in Embodiment 1 of the present invention.

[0024] Figure 2 This is a schematic diagram of the driving safety field model construction device proposed in Embodiment 2 of the present invention.

[0025] Figure 3This is a flowchart of the autonomous driving control method proposed in Embodiment 3 of the present invention.

[0026] Figure 4 This is a schematic diagram of the structure of the automatic driving control system proposed in Embodiment 4 of the present invention.

[0027] Figure 5 This is a schematic diagram of the time-domain evolution characteristics of the risk perception index in the simulated case of Embodiment 4 of the present invention.

[0028] Figure 6 This is a schematic diagram of the linear mapping calibration of the potential energy field and the physical reference in the simulation case of Embodiment 4 of the present invention.

[0029] Figure 7 This is a schematic diagram illustrating an example of updating risk thresholds based on environmental perception. Detailed Implementation

[0030] The specific embodiments of the present invention will now be described with reference to the accompanying drawings.

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the device proposed by this invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of this invention will become clearer from the following description. It should be noted that the drawings are in a very simplified form and use non-precise proportions, only for the purpose of conveniently and clearly illustrating the embodiments of this invention. Please refer to the accompanying drawings to make the objectives, features, and advantages of this invention more apparent and understandable. It should be understood that the structures, proportions, sizes, etc., depicted in the accompanying drawings are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed in the specification, and are not intended to limit the implementation conditions of this invention. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportional relationships, or adjustments to the size, without affecting the effects and objectives achieved by this invention, should still fall within the scope of the technical content disclosed in this invention.

[0032] Example 1 Figure 1 This is a flowchart of the method for constructing a driving safety field model proposed in Embodiment 1 of the present invention. Figure 1 As shown, the method for constructing a driving safety field model in this embodiment includes the following steps: Construct the dynamic field generated by a single obstacle, and calculate the sum of the dynamic fields generated by all obstacles within a preset area around the vehicle, as the risk potential energy of the vehicle; Establish a mapping equation between risk potential energy and the reciprocal of collision time; Obtain historical driving trajectory data of your own vehicle or vehicles of the same model as your own vehicle; Based on historical driving trajectory data, with the objective function of maximizing the linear fitting integration degree between risk potential energy and the reciprocal of collision time, the parameters in the mapping equation are iteratively searched and optimized to obtain the optimal parameter set. Substitute the optimal parameter set into the mapping equation to determine the driving safety field model; Collect real-time environmental parameters and update the risk threshold of the driving safety field model based on the real-time environmental parameters.

[0033] The driving safety field model constructed by the method in this embodiment belongs to the field model based on heterogeneous feature normalization. It can solve the equivalence problem of different traffic participants, such as pedestrians and trucks, under the same field model, and eliminate the failure of traditional indicators in low-speed and stationary states. The preset area around the vehicle can be defined by setting an influence radius, automatically calculating all obstacles within the radius, and can also be further constrained by additional conditions such as checking that the collision time is less than a preset number of seconds. After the safety field model is established, the optimal parameter set of the model is searched through iterative fitting of goodness of fit. Historical driving trajectory data includes motion feature information of the vehicle and obstacles, as well as the relative speed, relative position, and collision time of the vehicle and obstacles during driving, which are further calculated based on the motion feature information. Among them, the reciprocal of the collision time is the ratio of relative speed to relative position.

[0034] The method for constructing the driving safety field model according to this embodiment of the invention is further explained below.

[0035] 1. Construction of a dynamic field (dynamic target field) model based on heterogeneous feature normalization 1.1 Basic Model of Dynamic Fields Dynamic field of a single obstacle Defined as a multivariate linear function of distance, relative velocity, and equivalent mass coefficient. Specifically expressed as equation (1): (1) , They are respectively the bicycle and the first Euclidean distances and relative velocities between obstacles Indicates the first The field strength ratio coefficient of each obstacle.

[0036] (2) Let these be the coordinates of the vehicle. For the first The coordinates of the obstacle.

[0037] The target equivalent mass coefficient of the obstacle is determined by comprehensively considering the object's physical mass and geometric profile. For the car and the first The equivalent mass coefficient of each obstacle. For speed sensitivity index, This is the power of distance decay.

[0038] Furthermore, the sum of the dynamic fields generated by all obstacles within a preset area around the vehicle is taken as the risk potential energy of the vehicle. , expressed as equation (3): (3) in, is the anisotropic correction function used to describe the extension of risk in the direction of travel, where N represents the number of obstacles. Calculated using equation (4): (4) in, The longitudinal sensitivity coefficient, The relative azimuth angle is the angle between the target heading angle and the line connecting the measuring point, i.e., the first angle. The angle between the direction of motion (i.e., the velocity vector direction) of an obstacle (such as a pedestrian or other vehicle) and the line connecting the center point of the obstacle to the center point of the autonomous vehicle. This parameter, introduced as a baseline reference length, aims to perform dimensionless processing and numerical scaling of the kinematic variables within the exponential term.

[0039] 1.2 Physical baseline linear mapping equation Associating the linear transformation with the reciprocal of the collision time, the mapping equation is expressed as equation (5): (5) in, The slope of the mapping. This is a static risk bias. The mapping equation ensures that the field potential energy physically represents the average collision risk per unit time. This gives the risk potential energy a clear physical meaning. TTC represents Time-to-Collision, and the reciprocal of the collision time is the ratio of relative velocity to relative distance, i.e. .

[0040] 2. Data-driven calibration algorithm based on linear mapping 2.1 Constructing the calibration objective function The parameter set is calibrated using the following procedure: System data collection Group of natural driving data samples Construct the least squares optimization objective function according to equation (6). ,in: (6) in, For sample index, The total number of samples, A parameter smoothing penalty term (regularization) is used to prevent the model from overfitting at specific vehicle speeds. To control the penalty intensity coefficient. The parameter vector to be calibrated includes the unknown parameters to be calibrated in formula (5). .

[0041] 2.2 Design evaluation indicators and verify regression consistency. The calibration effect is obtained through goodness of fit. Quantified, defined as: (7) in, This represents the actual observed value of the risk potential energy. This is the calibrated predicted value of risk potential energy. This is the calibrated average risk potential energy. Only when... Only then is the parameter set considered to have engineering application value.

[0042] 2.3 Iterative Search Logic In the Particle Swarm Optimization (PSO) algorithm, the position vectors of the particles constitute the parameter set. The iteration update speed is determined by equation (8). and location : (8) in, Indicates the weighting coefficient. , Represents the learning factor. , This represents a random function. Let d be the d-th component of the individual extremum of the position vector of particle i in the t-th iteration. Let d represent the d-th dimension component of the global optimal solution for the position vector of particle i in the t-th iteration.

[0043] This process automates the transition from experience-based parameter tuning to data alignment.

[0044] This invention defines the dynamic field of obstacles based on relative velocity and relative displacement, eliminating the need for obstacle static / dynamic classification and simplifying the algorithm logic of the driving safety field model, thus eliminating the need for mode switching. When defining the risk potential energy of the vehicle, an anisotropic correction function is introduced to match the risk potential energy perception with the actual vehicle motion constraints, thereby generating a more realistic and reliable safety field model. Furthermore, the risk potential energy defined in this invention establishes a mathematical correspondence with the industry-recognized safety indicator of collision risk, giving the abstract potential energy a clear physical meaning. The potential energy gradient can accurately reflect the physical urgency of a collision, and it facilitates efficient calibration of model parameters using data-driven methods.

[0045] 3. Evolution and switching logic of risk thresholds in environmental perception For example, if the real-time environmental parameter is the road surface adhesion coefficient fed back by the sensor, the risk threshold can be corrected in real time according to equation (9): (9) in, The road surface adhesion coefficient, As a risk threshold, This is the revised risk threshold. This is the risk threshold before correction. It is an environmentally sensitive factor.

[0046] Furthermore, in this invention, the driver's driving style can also be considered when dynamically adjusting the risk threshold.

[0047] Example 2 Figure 2 This is a schematic diagram of the driving safety field model construction device proposed in Embodiment 2 of the present invention. Figure 2 As shown, the construction apparatus of this embodiment includes: The risk potential energy construction module is configured to construct the dynamic field generated by a single obstacle and calculate the sum of the dynamic fields generated by all obstacles within a preset area around the vehicle as the risk potential energy of the vehicle. The linear mapping construction module is configured to establish the mapping equation between the risk potential energy and the reciprocal of the collision time; The data acquisition module is configured to acquire historical driving trajectory data of the vehicle itself or vehicles of the same model as the vehicle itself, as well as acquire real-time environmental parameters; The parameter co-optimization module is configured to iteratively search and optimize the parameters in the mapping equation based on the historical driving trajectory data, with the objective function being to maximize the linear fitting integration degree between risk potential energy and the reciprocal of collision time, to obtain the optimal parameter set. The driving safety field calculation module is configured to substitute the optimal parameters into the mapping equation to determine the driving safety field model; The threshold evolution and risk cost map generation module is configured to update the risk threshold based on the real-time environmental parameters and generate a continuous risk cost map for the entire scenario based on the optimal parameter set.

[0048] Example 3 Figure 3 This is a flowchart of the automatic driving control method proposed in Embodiment 3 of the present invention. Figure 3 As shown, the autonomous driving control method of this embodiment includes the following steps: Collect real-time environmental parameters, monitor the vehicle's status, and perceive the motion characteristics of the vehicle and dynamic obstacles within the surrounding preset area; Calculate the reciprocal of the collision time based on the motion characteristic information; The motion feature information and the calculated reciprocal of the collision time are input into the pre-built autonomous driving safety field model to calculate the risk potential energy of the vehicle. When the risk potential energy is greater than or equal to the preset risk threshold, automatic emergency braking or forward collision warning is triggered. The driving safety field model is constructed using the following method: Define instantaneous risk potential energy, establish a mapping equation between the instantaneous risk potential energy and the reciprocal of the collision time, and based on historical driving trajectory data, take maximizing the linear fitting integration degree between risk potential energy and the reciprocal of the collision time as the objective function, iteratively search and optimize the parameters in the mapping equation to obtain the optimal parameter set, and substitute the optimal parameter set into the mapping equation to determine the driving safety field model. Based on the real-time environmental parameters, update the risk threshold; based on the optimal parameter set, generate a continuous risk cost map for the entire scenario.

[0049] Example 4 Figure 4 This is a schematic diagram of the structure of the automatic driving control system proposed in Embodiment 4 of the present invention. Figure 4 As shown, the autonomous driving control system in this embodiment involves five levels, L1 to L5, specifically including: The environmental perception and status acquisition layer (L1) is configured to collect real-time environmental parameters, monitor the vehicle's status, and perceive the motion characteristics of the vehicle and dynamic obstacles in the surrounding preset area. The kinematic evaluation index calculation layer (L2) is configured to calculate the reciprocal of the collision time based on the motion feature information. The calibration layer (L3) is configured to define risk potential energy, establish a mapping equation between the risk potential energy and the reciprocal of the collision time, and based on historical driving trajectory data, iteratively search and optimize the parameters in the mapping equation with the objective function of maximizing the linear fitting integration degree between the risk potential energy and the reciprocal of the collision time, to obtain the optimal parameter set. The optimal parameter set is then substituted into the mapping equation to determine the driving safety field model. The threshold evolution and risk cost map generation layer (L4) is configured to update the risk threshold based on the real-time environmental parameters and generate a continuous risk cost map for the entire scenario based on the optimal parameter set output by the calibration layer. The decision planning and active safety application layer (L5) is configured to input the calculated reciprocal of the collision time into the autonomous driving safety field model, output the risk potential energy of the vehicle, and determine whether the risk potential energy is greater than or equal to the risk threshold. If so, it triggers automatic emergency braking or forward collision warning; otherwise, it performs path planning based on the risk cost map and searches for a potential energy depression path.

[0050] Case Study: This case study simulates a typical emergency following scenario on a highway.

[0051] The main vehicle (Ego) travels at a constant speed of 33.3 m / s (approximately 120 km / h), with an initial following distance of 60 m. At that moment, the vehicle in front suddenly applied emergency braking (deceleration of 5m / s). ).

[0052] Independent variable: Relative distance changing over time and relative velocity .

[0053] Dependent variables: Physical baseline index 1 / TTC and risk potential .

[0054] 1) Mapping accuracy analysis Regression analysis was performed on the 100 sets of spatiotemporal trajectory sampling points generated, and the results are as follows: Linear correlation: Figure 5 This is a schematic diagram of the time-domain evolution characteristics of the risk perception index in the simulated case of Embodiment 4 of the present invention. Figure 6 This is a schematic diagram illustrating the linear mapping calibration of the potential energy field and the physical reference in the simulation example of Embodiment 4 of the present invention. Figure 5 , Figure 6 As shown, the linear mapping verification results between risk potential energy and the reciprocal of TTC demonstrate the calibrated risk potential energy. The physical index 1 / TTC exhibits extremely high linearity across the entire time axis, with a goodness of fit of [missing information]. It reached 0.9.

[0055] Slope stability: Mapped slope During periods of severe risk fluctuations ( to The parameters remained stable, with no significant drift observed.

[0056] Therefore, it is fully demonstrated that the parameter collaborative optimization algorithm proposed in the embodiments of the present invention can accurately capture the essence of physical risk and successfully quantify the abstract field potential energy into a collision urgency with physicality.

[0057] 2) Environmental Adaptability Analysis: Table 1 shows the decision-making performance under different road surface adhesion coefficients, verifying the dynamic evolution capability of the system.

[0058] Table 1

[0059] Traditional TTC (Traffic Tightness Control) indicators typically use fixed thresholds, failing to detect the increased braking distance caused by slippery road surfaces. As shown in Table 1, this embodiment of the invention uses a dynamic threshold evolution module to automatically lower the safety boundary when an increased environmental risk is detected. Under slippery road conditions, the system can trigger a warning approximately 1 second in advance, which is equivalent to an additional 30 meters of avoidance space during high-speed driving, greatly reducing the risk of rear-end collisions.

[0060] 3) Computational stability and continuity analysis Singularity Suppression: Traditional TTC metrics in relative velocity 0 (e.g., constant speed following) or relative distance At a relative speed of 0, numerical mutations or failures are likely to occur. While maintaining a high degree of consistency with classical kinematic indices, this invention can solve the singular value failure problem of TTC when the relative speed is zero. It can provide the autonomous driving decision-making and planning layer with a continuous risk cost map in the entire time domain and all directions, improving the safety and driving comfort of vehicles in complex interaction scenarios such as lane changing, emergency following, etc.

[0061] Smoothness: Figure 7 This is a schematic diagram illustrating an example of updating risk thresholds based on environmental perception. (Observe...) Figure 7 As you can see, the output of this example is The curve was smooth throughout the braking process, without any abrupt changes.

[0062] Engineering Value: The continuously differentiable risk potential energy curve designed in this embodiment can be directly used as the cost function of the underlying autonomous driving system. Compared with discrete TTC triggering, the continuous potential energy field makes the actuators (such as brake pumps and steering motors) move more linearly, avoiding frequent "nodding" or swaying of the vehicle during extreme obstacle avoidance, thus improving driving comfort. This embodiment solves the pain point of discrete failure of traditional kinematic indicators through continuous and smooth output.

[0063] The driving safety field construction method proposed in this embodiment of the invention has a goodness of fit. A value >0.9 ensures the scientific rigor of the risk assessment and achieves precise alignment with physical benchmarks. The dynamic threshold mechanism significantly enhances safety under severe weather conditions and possesses environmental awareness capabilities. It meets real-time control requirements and addresses the discrete failure issue of traditional kinematic indicators through continuous and smooth output.

[0064] Another embodiment of the present invention provides an electronic terminal, including a processor and a memory, wherein the memory stores a computer program, and the processor calls the computer program to execute: the steps of the construction method described in Embodiment 1 or the steps of the control method described in Embodiment 3.

[0065] Another embodiment of the present invention provides a readable storage medium storing a computer program that, when invoked by a processor, executes the steps of the construction method described in Embodiment 1 or the steps of the control method described in Embodiment 3.

[0066] Another embodiment of the present invention provides a computer program product, including a computer program / instruction, characterized in that: when the computer program / instruction is executed by a processor, it implements the steps of the construction method described in Embodiment 1 or the steps of the control method described in Embodiment 3.

[0067] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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, they should be considered to be within the scope of this specification.

[0068] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for constructing a driving safety field model, characterized in that, Including the following steps: Define the dynamic field generated by a single obstacle, and calculate the sum of the dynamic fields generated by all obstacles within a preset area around the vehicle, as the risk potential energy of the vehicle. Establish a mapping equation between the risk potential energy and the reciprocal of the collision time; Obtain historical driving trajectory data of your own vehicle or vehicles of the same model as your own vehicle; Based on the historical driving trajectory data, with the optimization objective of maximizing the linear fitting integration degree between risk potential energy and the reciprocal of collision time, the parameters in the mapping equation are iteratively searched and optimized to obtain the optimal parameter set. Substitute the optimal parameter set into the mapping equation to determine the driving safety field model; Collect real-time environmental parameters and update the risk threshold of the driving safety field model based on the real-time environmental parameters; The dynamic field of the single obstacle is represented by equation (1): （1） in, Represents the dynamic field of a single obstacle; , They are respectively the bicycle and the first Euclidean distance and relative velocity between obstacles; Indicates the first The field strength ratio coefficient of each obstacle; Indicates the first The angle between the heading angle of the obstacle and the line connecting the vehicle; The equivalent mass coefficient of the obstacle. For the car and the first The equivalent mass coefficient of each obstacle; For speed sensitivity index, The distance decay is a power of 1. For anisotropy correction function; The risk potential energy is expressed as equation (3): （3） in, N represents the vehicle's potential energy at risk; N represents the number of obstacles. Calculated using equation (4): （4） in, Here is the directional anisotropy coefficient. Used as a reference length; The mapping equation is expressed as equation (5): （5） in, The slope of the mapping. This is a static risk bias; It is the countdown to the collision time; The historical driving trajectory data includes the relative speed, relative position, and collision time between the vehicle and the obstacle, with the reciprocal of the collision time being the ratio of the relative speed to the relative position. The parameter set is calibrated using the following procedure: collection Group of natural driving data samples Construct the least squares optimization objective function according to equation (6). : （6） in, For sample index, The total number of samples, For parameter smoothing penalty term; To control the penalty intensity coefficient; The parameter vector to be calibrated, including the unknown parameters to be calibrated. ; The particle swarm optimization algorithm is used to optimize the objective function. The parameters to be determined are iteratively optimized.

2. The method for constructing a driving safety field model as described in claim 1, characterized in that, The real-time environmental parameter is the road surface adhesion coefficient fed back by the sensor, and the risk threshold is corrected in real time according to equation (9): （9） in, The road surface adhesion coefficient, As a risk threshold, This is the revised risk threshold. This is the risk threshold before correction. It is an environmentally sensitive factor.

3. A driving safety field model construction apparatus, executing the driving safety field model construction method according to any one of claims 1-2, characterized in that, The device includes: The risk potential energy construction module is configured to construct the dynamic field generated by a single obstacle and calculate the sum of the dynamic fields generated by all obstacles within a preset area around the vehicle as the risk potential energy of the vehicle. The linear mapping construction module is configured to establish the mapping equation between the risk potential energy and the reciprocal of the collision time; The data acquisition module is configured to acquire historical driving trajectory data of the vehicle itself or vehicles of the same model as the vehicle itself, as well as acquire real-time environmental parameters; The parameter co-optimization module is configured to, based on the historical driving trajectory data, take maximizing the linear fitting integration degree between risk potential energy and the reciprocal of collision time as the optimization objective, iteratively search and optimize the parameters in the mapping equation to obtain the optimal parameter set. The driving safety field calculation module is configured to substitute the optimal parameters into the mapping equation to determine the driving safety field model; The threshold evolution and risk cost map generation module is configured to update the risk threshold based on the real-time environmental parameters and generate a continuous risk cost map for the entire scenario based on the optimal parameter set.

4. An automatic driving control method, characterized in that, Including the following steps: Collect real-time environmental parameters, monitor the vehicle's status, and perceive the motion characteristics of the vehicle and dynamic obstacles within the surrounding preset area; Calculate the reciprocal of the collision time based on the motion characteristic information; The motion feature information and the calculated reciprocal of the collision time are input into the pre-built autonomous driving safety field model to calculate the risk potential energy of the vehicle. When the risk potential energy is greater than or equal to the preset risk threshold, automatic emergency braking or forward collision warning is triggered. The driving safety field model is constructed using the driving safety field model construction method according to any one of claims 1-2, including the following steps: Define instantaneous risk potential energy, establish a mapping equation between the instantaneous risk potential energy and the reciprocal of the collision time, and based on historical driving trajectory data, take maximizing the linear fitting integration degree between risk potential energy and the reciprocal of the collision time as the optimization objective, iteratively search and optimize the parameters in the mapping equation to obtain the optimal parameter set, and substitute the optimal parameter set into the mapping equation to determine the driving safety field model. Based on the real-time environmental parameters, update the risk threshold; based on the optimal parameter set, generate a continuous risk cost map for the entire scenario.

5. An automatic driving control system, characterized in that, include: The environmental perception and status acquisition layer is configured to collect real-time environmental parameters, monitor the vehicle's status, and perceive the motion characteristics of the vehicle and dynamic obstacles within the surrounding preset area. The kinematic evaluation index calculation layer is configured to calculate the reciprocal of the collision time based on the motion characteristic information. The calibration layer is configured to define risk potential energy, establish a mapping equation between the risk potential energy and the reciprocal of the collision time, and based on historical driving trajectory data, with the optimization objective of maximizing the linear fitting integration degree between the risk potential energy and the reciprocal of the collision time, iteratively search and optimize the parameters in the mapping equation to obtain the optimal parameter set. The optimal parameter set is then substituted into the mapping equation to determine the driving safety field model. The threshold evolution and risk cost graph generation layer is configured to update the risk threshold based on the real-time environmental parameters; Based on the optimal parameter set output by the calibration layer, a continuous risk cost map for the entire scenario is generated; The decision planning and active safety application layer is configured to input the calculated reciprocal of the collision time into the autonomous driving safety field model, output the risk potential energy of the vehicle, and determine whether the risk potential energy is greater than or equal to the risk threshold. If so, it triggers automatic emergency braking or forward collision warning; otherwise, it performs path planning based on the risk cost map and searches for a potential energy depression path. The driving safety field model is constructed using any one of the driving safety field model construction methods described in claims 1-2.

6. An electronic terminal, characterized in that: It includes a processor and a memory, the memory storing a computer program, the processor calling the computer program to perform the steps of the construction method according to any one of claims 1-2.

7. A readable storage medium, characterized in that: A computer program is stored, which, when invoked by a processor, executes the steps of the construction method according to any one of claims 1-2.

8. A computer program product comprising a computer program / instructions, characterized in that: When the computer program / instructions are executed by the processor, they implement the steps of the construction method according to any one of claims 1-2.

Images