Safety control methods, electronic devices, and storage media for vehicles

Abstract

This application provides a safety control method, electronic device, and storage medium for a vehicle. The safety control method is applied to a target vehicle, which includes a primary driving system and a backup driving system. The safety control method includes: a control acquisition module collecting environmental data during the vehicle's operation and transmitting the environmental data to a monitoring module, so that the monitoring module can determine whether the target vehicle is in a warning state based on the environmental data; if the target vehicle is in a warning state, the primary driving system is controlled to enter a pre-switching mode; the control monitoring module monitors the vehicle's operating status and determines whether the primary driving system has malfunctioned; if so, the vehicle's control system is switched from the primary driving system to the backup driving system. The technical problem solved by this application is that the decision-making ability of the existing dual driving systems is insufficient, resulting in a long reaction time when the primary driving system switches to the backup driving system, during which the vehicle is prone to loss of control.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and more specifically, to a safety control method, electronic device, and storage medium for vehicles. Background Technology

[0002] With the rapid development of autonomous driving technology and advanced driver assistance systems, vehicle safety and reliability have become key concerns in the industry. Modern vehicles are typically equipped with multiple sensors to perceive their surroundings and achieve autonomous or assisted driving functions through complex decision-making algorithms. To further enhance the fault tolerance of autonomous driving, vehicles are equipped with dual driving systems, which can switch to a backup driving system in case of failure of the primary driving system.

[0003] However, the relevant technology has at least one of the following problems: the decision-making ability of the existing dual driving system is insufficient, resulting in a long reaction time when the primary driving system switches to the backup driving system, during which the vehicle is prone to loss of control. Summary of the Invention

[0004] The technical problem solved by this invention is that the decision-making ability of the existing dual-driving system is insufficient, resulting in a long reaction time when the primary driving system switches to the backup driving system, during which the vehicle is prone to loss of control.

[0005] To address the aforementioned problems, this invention provides a safety control method for vehicles. The method is applied to a target vehicle, which includes a primary driving system and a backup driving system. The safety control method includes: controlling a data acquisition module of the target vehicle to collect environmental data during the vehicle's operation and transmitting the environmental data to a monitoring module, so that the monitoring module can determine whether the target vehicle is in a warning state based on the environmental data; if the target vehicle is in a warning state, controlling the primary driving system to enter a pre-switching mode to prepare for switching from the primary driving system to the passenger driving system; controlling the monitoring module to monitor the operating status of the target vehicle and determine whether the primary driving system has malfunctioned; if so, controlling the vehicle's control system to switch from the primary driving system to the backup driving system.

[0006] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by analyzing the environment of the target vehicle through the surrounding environmental data, it can be determined whether the target vehicle is in a pre-warning state prone to failure, thereby activating the pre-switching mode of the main driving system to prepare for possible system switching and reduce reaction time; when the main driving system fails, the control system of the target vehicle is switched from the main driving system to the backup driving system, thereby ensuring the continuity of the target vehicle's driving in failure scenarios.

[0007] In one embodiment of the present invention, to enable the monitoring module to determine whether the target vehicle is in a warning state based on environmental data, the method includes: calculating the uncertainty parameter σ of the driver's system and the dynamic uncertainty threshold σ based on the environmental data. th Determine whether the uncertainty parameter σ is greater than the dynamic uncertainty threshold σ. th If yes, the vehicle is determined to be in a warning state; otherwise, the vehicle is determined to be controlled by the driver's system.

[0008] Compared with existing technologies, the technical effects achieved by this solution are as follows: by calculating the uncertainty parameter σ and the dynamic uncertainty threshold σ th This quantifies the operational risks of the driver's system; only when the operational risks reach a certain level is the pre-switching mode of the driver's system activated, reducing unnecessary system operations, lowering the misjudgment rate, and thus optimizing the system's operational efficiency.

[0009] In one embodiment of the present invention, environmental data includes traffic density D, vehicle speed change rate V, weather conditions T, and road characteristics W; the uncertainty parameter σ and dynamic uncertainty threshold σ of the main driving system are calculated based on the environmental data. th This includes: calculating environmental complexity C based on traffic density D, vehicle speed change rate V, weather conditions T, and road characteristics W; and calculating the dynamic uncertainty threshold σ based on environmental complexity C. th And follow the formula 3 below:

[0010] Formula 3: ;

[0011] Where, σ base This serves as a baseline uncertainty threshold, used to replace the dynamic uncertainty threshold σ in simple scenarios. th Use, P human Let σ be the probability that the driver takes over the target vehicle. total The total uncertainty of multimodal fusion is represented by k1, k2, and α, which are proportionality coefficients.

[0012] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by calculating the complexity of the environment, comprehensively considering the environmental characteristics of the target vehicle, and adaptively adjusting the operational risks of the target vehicle when facing different environmental characteristics, it not only ensures the efficiency of the main driving system during normal operation, but also enhances the safety guarantee of the target vehicle when facing abnormal situations.

[0013] In one embodiment of the present invention, the primary driving system employs a first decision algorithm to handle driving tasks under normal circumstances; the backup driving system employs a second decision algorithm to handle driving tasks under emergency circumstances; wherein, the predictability of the second decision algorithm is higher than that of the first decision algorithm.

[0014] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by using a first decision algorithm and a second decision algorithm to handle different tasks in the primary driving system and the backup driving system, the overlap between the primary driving system and the backup driving system is reduced, and the possibility of the primary driving system and the backup driving system failing for the same reason is reduced; in addition, by improving the predictability of the second decision algorithm, the safety of the target vehicle when it is controlled by the backup driving system is improved.

[0015] In one embodiment of the present invention, the acquisition module includes a lidar, a camera, a millimeter-wave radar, and an ultrasonic radar; the acquisition module for controlling the target vehicle acquires environmental data during the vehicle's driving process, including: controlling the lidar to acquire point cloud data; controlling the camera to acquire image data; controlling the millimeter-wave radar to acquire target detection data; and controlling the ultrasonic radar to acquire near-field detection data.

[0016] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by collecting a variety of data through multiple sensors, the vehicle's perception of the surrounding environment is enriched, and the collaborative work of multiple sensors can compensate for the shortcomings of specific sensors under specific weather conditions, thereby improving the integrity and reliability of environmental data.

[0017] In one embodiment of the present invention, the acquisition module of the controlled vehicle acquires environmental data during the vehicle's driving process, including: calculating environmental data using a Kalman filter algorithm based on point cloud data, image data, target detection data, and near-field detection data; wherein the weight ratio of point cloud data, image data, target detection data, and near-field detection data in the Kalman filter algorithm is determined by the vehicle's position, speed, and obstacle information.

[0018] Compared with existing technologies, the technical effects achieved by adopting this solution are as follows: by fusing multimodal data through the Kalman filter algorithm, sensor noise interference is reduced, and the accuracy and consistency of environmental data are improved; in addition, the weights of each sensor data are dynamically adjusted based on vehicle position, speed and obstacle information, optimizing the applicability of data processing in different scenarios.

[0019] In one embodiment of the present invention, environmental data is calculated using a Kalman filter algorithm based on point cloud data, image data, target detection data, and near-field detection data, including: initializing the state vector x. k State vector x k This describes the vehicle's speed, position coordinates, and the number and position coordinates of obstacles in the environment; the state prediction step is performed using the state transition function f(·), resulting in the predicted state vector. and prediction error covariance matrix The update step involves calculating the observed value z of acquisition module s based on the data acquisition module s. s and the corresponding observation matrix H s According to the observed value z s and observation matrix H s The Kalman gain was calculated. ; Calculate the confidence level w of the acquisition module s s And follow the formula below:

[0020] ;

[0021] in, It is a measure of the uncertainty of the acquisition module s. It refers to the level of uncertainty. This is a sensitivity parameter used to adjust the steepness of the confidence curve; it is based on the confidence level w of the acquisition module s. s Kalman gain The adaptive Kalman gain is obtained by performing a weighted average. According to adaptive Kalman gain Predicted state vector and prediction error covariance matrix The updated state vector is calculated. and update the error covariance matrix .

[0022] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by introducing an adaptive weight adjustment mechanism based on the uncertainty of the acquisition module, the main driving system can automatically adjust the contribution ratio of each sensor in different driving environments, thereby improving the accuracy and robustness of the fusion results.

[0023] In one embodiment of the present invention, the target vehicle includes critical modules and non-critical modules, and the target vehicle has a higher priority in the allocation of computing power to the critical modules than to the critical modules.

[0024] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: priority is given to allocating computing power to key modules to ensure the stable operation of core functions under high load scenarios; non-key modules adopt time-sharing multiplexing to optimize the utilization of computing resources and reduce system power consumption and hardware costs.

[0025] On the other hand, the present invention also provides an electronic device, which includes: a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the security control method as in any of the above examples.

[0026] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: it can achieve the technical effects corresponding to any of the above examples, which will not be elaborated here.

[0027] On the other hand, the present invention also provides a storage medium on which a program or instruction is stored, which, when executed by a processor, implements the security control method as in any of the above examples.

[0028] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: it can achieve the technical effects corresponding to any of the above examples, which will not be elaborated here.

[0029] By adopting the technical solution of the present invention, the following technical effects can be achieved:

[0030] (1) Analyze the environment of the target vehicle by using the surrounding environmental data of the vehicle, and then determine whether the target vehicle is in a state of easy failure warning, thereby activating the pre-switching mode of the main driving system to prepare for possible system switching and reduce reaction time; when the main driving system fails, switch the control system of the target vehicle from the main driving system to the backup driving system to ensure the continuity of the target vehicle in the failure scenario.

[0031] (2) By calculating the uncertainty parameter σ and the dynamic uncertainty threshold σ th This quantifies the operational risks of the driver's system; only when the operational risks reach a certain level is the pre-switching mode of the driver's system activated, reducing unnecessary system operations, lowering the misjudgment rate, and thus optimizing the system's operational efficiency.

[0032] (3) By calculating the environmental complexity, the environmental characteristics of the target vehicle are comprehensively considered, and the operational risks of the target vehicle when facing different environmental characteristics are adaptively adjusted. This ensures the efficiency of the main driving system during normal operation and enhances the safety guarantee of the target vehicle when facing abnormal situations. Attached Figure Description

[0033] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a flowchart of a vehicle safety control method provided in an embodiment of the present invention. Detailed Implementation

[0035] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0036] See Figure 1 This is a flowchart illustrating a vehicle safety control method provided in an embodiment of the present invention. The present invention provides a vehicle safety control method, which is applied to a target vehicle, the target vehicle including a primary driving system and a backup driving system. The safety control method includes:

[0037] S100: The data acquisition module of the controlled vehicle collects environmental data during the vehicle's driving process and transmits the environmental data to the monitoring module so that the monitoring module can determine whether the vehicle is in a warning state based on the environmental data.

[0038] S200: If the target vehicle is in a warning state, control the driver system to enter a pre-switching mode to prepare to switch from the driver system to the passenger system;

[0039] S300: The control and monitoring module monitors the operating status of the target vehicle and determines whether the driver's system has malfunctioned.

[0040] S400: If so, the vehicle control system switches from the primary driving system to the backup driving system.

[0041] By analyzing the environment around the vehicle, the system can determine whether the vehicle is in a state of high risk of failure. This activates the pre-switch mode of the driver's system to prepare for a possible system switch and reduce reaction time. In the event of a failure in the driver's system, the system switches the vehicle's control system from the driver's system to the backup driver system, thereby ensuring the continuity and safety of the vehicle in failure scenarios.

[0042] Preferably, when the target vehicle is in adverse conditions such as dangerous weather or complex road conditions, the driver's system on the target vehicle is prone to failure. When the adverse conditions accumulate to a certain extent, an alarm can be issued to determine that the target vehicle is in a warning state.

[0043] Preferably, the pre-switch mode indicates that the vehicle is in a state of preparing to switch from the primary driving system to the backup driving system, so as to reduce the reaction time required for switching when the vehicle malfunctions.

[0044] Preferably, the monitoring module is located inside the vehicle to monitor the vehicle's internal operating status, and the data acquisition module is located around the vehicle to collect environmental data from the surrounding area.

[0045] Preferably, the vehicle described in this invention is an autonomous vehicle with at least two driving systems, and is normally driven autonomously by the vehicle itself.

[0046] Furthermore, S100 includes:

[0047] S110: Calculate the uncertainty parameter σ and dynamic uncertainty threshold σ of the main driving system based on environmental data. th Determine whether the uncertainty parameter σ is greater than the dynamic uncertainty threshold σ. th ;

[0048] S120: If so, the target vehicle is determined to be in a warning state;

[0049] S130: If not, then the vehicle is determined to be driven by the driver's system.

[0050] The prior art discloses a method and system for controlling the operation of an autonomous vehicle, in which the main driving system controls the vehicle's movement, and if a problem occurs in the vehicle's driving status, it switches to a backup driving system. In determining whether the main driving system is operating normally, it mainly collects the status signals emitted by the main driving system at predetermined time intervals to see if they are normal, or whether there is a malfunction in the vehicle's internal software and hardware. Both of these methods require a certain amount of reaction time, but the actual reaction time that can be reserved when a malfunction occurs is very small, and the reaction time needs to be further shortened.

[0051] By calculating the uncertainty parameter σ and the dynamic uncertainty threshold σ th It can quantify the operational risks of the driver's system; when the operational risks of the vehicle accumulate to a certain level, it will activate the pre-switching mode of the driver's system, and reduce the reaction space of the monitoring module when the driver's system has a problem.

[0052] Preferably, the master driving system employs an end-to-end autonomous driving algorithm based on a deep neural network (DNN), and calculates the uncertainty parameter σ and the dynamic uncertainty threshold σ based on an uncertainty quantification method. th The master control system can be modeled as a complex mapping function f. DNN This maps the input X from the acquisition module to the control output Y, following Formula 1:

[0053] Formula 1: ;

[0054] Here, θ represents the parameters of the deep neural network of the model. Furthermore, the uncertainty σ of the prediction result is estimated using Bayesian deep learning techniques. 2This transforms the model's output from a single value into a normal distribution, following Formula 2:

[0055] Formula 2: ;

[0056] Among them, when the uncertainty parameter σ exceeds the dynamic uncertainty threshold σ th At that time, we believed that the driver's system might be at risk of failure.

[0057] Furthermore, environmental data includes traffic density D, vehicle speed change rate V, weather conditions T, and road characteristics W; S110 includes:

[0058] S111: Calculate the environmental complexity C based on traffic density D, vehicle speed change rate V, weather conditions T, and road characteristics W;

[0059] S112: Calculate the dynamic uncertainty threshold σ of the master driving system based on the environmental complexity C. th And follow the formula 3 below:

[0060] Formula 3: ;

[0061] Where, σ base This serves as a baseline uncertainty threshold, used to replace the dynamic uncertainty threshold σ in simple scenarios. th Use, P human Let σ be the probability that the driver takes over the target vehicle. total Let α represent the total uncertainty of sensor multimodal fusion, and k1, k2, and α be proportionality coefficients.

[0062] By calculating the environmental complexity, the environmental characteristics of the target vehicle are comprehensively considered, and the operational risks of the target vehicle when facing different environmental characteristics are adaptively adjusted. This ensures both the efficiency of the main driving system during normal operation and enhances the safety of the target vehicle when facing abnormal situations.

[0063] Preferably, when calculating the environmental complexity C based on traffic density D, vehicle speed change rate V, weather conditions T, and road characteristics W, the following formula 4 is followed:

[0064] Formula 4: ;

[0065] Preferably, w1, w2, w3, and w4 are weighting coefficients.

[0066] Preferably, the three proportionality coefficients k1, k2, and α are used to adjust the environmental complexity, total uncertainty, and driver behavior for the dynamic uncertainty threshold σ, respectively. th The impact.

[0067] Specifically, this invention provides a method for handling real-world scenarios where the uncertainty parameter σ is greater than the dynamic uncertainty threshold σ. th Example 1. The specific scenario is that the vehicle is driving on a mountain highway in foggy weather and encounters a sudden traffic jam. The specific values ​​of each parameter in this scenario can be seen in Table 1.

[0068] Table 1. Values ​​of various parameters of the object vehicle when driving in the scenario of Example 1.

[0069] Specifically, in Example 1 scenario, the traffic density D is 0.92, indicating extremely high density (where the traffic density D ranges from 0 to 1, with 1 indicating complete congestion), the vehicle speed change rate V is 0.85 (indicating frequent sudden braking, with a value range of 0 to 1, and 1 indicating drastic changes), the weather condition T is 0.95 (indicating dense fog with visibility <50 meters), and the road feature W is 0.88 (indicating continuous steep slopes and sharp bends).

[0070] Specifically, in the scenario of Example 1, the model output is obtained as follows: Y = 0.18 rad and uncertainty parameter σ = 0.26 rad can be measured through Bayesian deep learning (Monte Carlo Dropout sampling 200 times).

[0071] Specifically, by substituting traffic density D, vehicle speed change rate V, weather conditions T, and road characteristics W into Formula 4, we can obtain that the environmental complexity C = 0.9075.

[0072] Specifically, the environmental complexity C and the baseline uncertainty threshold σ are considered. base Driver takeover probability P human And the total uncertainty σ of sensor multimodal fusion total Substituting into Formula 3, we can obtain the dynamic uncertainty threshold σ. th The value is 0.0298 rad; 0.26 rad > 0.0298 rad, therefore the uncertainty parameter σ is greater than the dynamic uncertainty threshold σ. th The vehicle in question is determined to be in a warning state.

[0073] Specifically, the meanings and effects of each factor in Example 1 are as follows: (1-k1·C) is the environmental complexity term, with a value of 0.36475. Due to the high complexity of the environment, the dynamic uncertainty threshold σ is increased. th Compressed to the baseline uncertainty threshold σ base 36.5%;

[0074] (1+α·P) human The value σ represents the driver takeover item, set to 1.125. It is influenced by the probability of driver takeover and represents a dynamic uncertainty threshold. thProvides a small buffer;

[0075] exp(-k2·σ total The multimodal fusion term, with a value of approximately 0.6042, represents the multimodal fusion term. Due to interference from dense fog, the uncertainty of the sensor's multimodal fusion term is relatively high, further suppressing the dynamic uncertainty threshold σ. th .

[0076] Preferably, when it is determined that the target vehicle is in a warning state, the audible and visual alarm on the target vehicle is activated to prompt the driver to take over, and the speed of the target vehicle is limited to 40km / h.

[0077] Furthermore, the primary driving system employs a first decision algorithm to handle driving tasks under normal circumstances; the backup driving system employs a second decision algorithm to handle driving tasks under emergency circumstances; the second decision algorithm has higher predictability than the first decision algorithm.

[0078] By employing a first decision algorithm and a second decision algorithm that handle different tasks in the primary driving system and the backup driving system, the overlap between the primary driving system and the backup driving system is reduced, and the possibility of the primary driving system and the backup driving system failing for the same reason is lowered. In addition, by improving the predictability of the second decision algorithm, the safety of the target vehicle when it is controlled by the backup driving system is improved.

[0079] Preferably, the backup driving system employs a rule-based traditional decision-making algorithm, typically following a set of predefined logical rules R to make decisions. Assuming these rules are deterministic, then we have Formula 5:

[0080] Formula 5: ;

[0081] Among them, the function g(·) is a function composed of a series of conditional statements formulated according to traffic regulations and safety standards, such as the MPC (Model Predictive Control) control algorithm. Because this method does not involve complex parameter optimization, its behavior is more predictable and easier to understand.

[0082] Furthermore, the acquisition module includes a lidar, a camera, a millimeter-wave radar, and an ultrasonic radar; the acquisition module for the controlled vehicle acquires environmental data during the vehicle's driving process, including: controlling the lidar to acquire point cloud data; controlling the camera to acquire image data; controlling the millimeter-wave radar to acquire target detection data; and controlling the ultrasonic radar to acquire near-field detection data.

[0083] By collecting various data through multiple sensors, the vehicle's perception of its surrounding environment is enriched. Furthermore, the collaborative work of multiple sensors can compensate for the shortcomings of specific sensors under specific weather conditions, thereby improving the integrity and reliability of environmental data.

[0084] Preferably, the lidar model is Velodyne VLP-16 with a horizontal field of view of 360 degrees and a vertical field of view of 30 degrees; the camera is a 1080P high-definition camera; the millimeter-wave radar is Continental ARS408 with a working frequency of 77GHz; and the ultrasonic radar is Continental 3TX.

[0085] Furthermore, the acquisition module of the controlled vehicle collects environmental data during the vehicle's driving process, including: calculating environmental data using a Kalman filter algorithm based on point cloud data, image data, target detection data, and near-field detection data; wherein, the weight ratio of point cloud data, image data, target detection data, and near-field detection data in the Kalman filter algorithm is determined by the vehicle's position, speed, and obstacle information.

[0086] By fusing multimodal data using the Kalman filter algorithm, sensor noise interference is reduced, and the accuracy and consistency of environmental data are improved. In addition, the weights of each sensor data are dynamically adjusted based on vehicle position, speed, and obstacle information to optimize the applicability of data processing in different scenarios.

[0087] Furthermore, environmental data is calculated using the Kalman filter algorithm based on point cloud data, image data, target detection data, and near-field detection data, including: initializing the state vector x. k State vector x k This is used to describe the speed and position coordinates of the vehicle, as well as the number and position coordinates of obstacles in the environment, and follows the formula 6 below:

[0088] Formula 6: ;

[0089] Where, p x p y This indicates the vehicle's position on a two-dimensional plane; v x v y Represents the vehicle's velocity component; o xi o yi Let represent the position coordinates of the i-th obstacle; n is the number of obstacles in the environment; T represents the transpose of the matrix; the state prediction step is performed using the state transition function f(·) to obtain the predicted state vector. and prediction error covariance matrix And follow the formulas 7-8 below:

[0090] Formula 7: ;

[0091] Formula 8: ;

[0092] in, It is the predicted state vector; It is the prediction error covariance matrix; and It is a priori state estimate, represented as a predicted value, without incorporating current measurements; and It is the posterior state estimate, represented as the updated value, which incorporates the current measurement; f(·) refers to the state transition function; It is a control input; It is a Jacobian matrix; This refers to the process noise covariance matrix; k represents the current time, and k-1 represents the previous time.

[0093] The update step involves calculating the observed value z of acquisition module s based on the data acquisition module s. s and the corresponding observation matrix H s According to the observed value z s and observation matrix H s The Kalman gain was calculated. Then, based on the Kalman gain Predicted state vector and prediction error covariance matrix The updated state vector is calculated. and update the error covariance matrix And follow the formulas 9-13 below:

[0094] Formula 9: ;

[0095] Formula 10: ;

[0096] Formula 11: ;

[0097] Formula 12: ;

[0098] Formula 13: ;

[0099] in, It is the observation noise covariance matrix. It is the Kalman gain, h s (·) is the observation function. It measures the noise covariance matrix; it calculates the confidence level w of the acquisition module s. s And follow the formula 14 below:

[0100] Formula 14: ;

[0101] in, It is a measure of the uncertainty of the acquisition module s. It refers to the level of uncertainty. It is a sensitivity parameter used to adjust the steepness of the confidence curve;

[0102] In the update step, the confidence level w of the acquisition module s is used as a reference. s Kalman gain The adaptive Kalman gain is obtained by performing a weighted average. According to adaptive Kalman gain Predicted state vector and prediction error covariance matrix The updated state vector is calculated. and update the error covariance matrix And follow the formulas 15-17 below:

[0103] Formula 15: ;

[0104] Formula 16: ;

[0105] Formula 17: ;

[0106] Where z represents all observations The combination of h(·) is the comprehensive observation function.

[0107] By introducing an adaptive weight adjustment mechanism based on the uncertainty of the acquisition module, the main driving system can automatically adjust the contribution ratio of each sensor in different driving environments, thereby improving the accuracy and robustness of the fusion results.

[0108] Preferably, in the acquisition module s, lider represents a laser radar; camera represents a camera; radar represents a millimeter-wave radar; and ultrasonic represents an ultrasonic radar.

[0109] Furthermore, the target vehicle includes critical modules and non-critical modules, and the computing power allocation for critical modules in the target vehicle has a higher priority than that for critical modules.

[0110] Prioritize allocating computing power to critical modules to ensure the stable operation of core functions under high load scenarios; non-critical modules adopt time-sharing multiplexing to optimize the utilization of computing resources and reduce system power consumption and hardware costs.

[0111] Preferably, the key modules include an environmental perception module, a decision-making and planning module, and a control execution module, while the non-key modules include a human-machine interaction module, an in-vehicle multimedia module, etc.

[0112] Preferably, the critical modules run on an eight-core ARM processor with a main frequency of 3.0 GHz to process sensor data and make decision instructions in real time; the non-critical modules run on a quad-core ARM processor with a main frequency of 2.0 GHz and are time-division multiplexed using a priority-based preemptive scheduling algorithm.

[0113] On the other hand, the present invention also provides an electronic device, which includes: a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the security control method as in any of the above examples.

[0114] The technical effects that can be achieved in any of the above examples will not be elaborated here.

[0115] On the other hand, the present invention also provides a storage medium on which a program or instruction is stored, which, when executed by a processor, implements the security control method as in any of the above examples.

[0116] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: it can achieve the technical effects corresponding to any of the above examples, which will not be elaborated here.

[0117] While the present invention has been disclosed above, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A safety control method for vehicles, characterized in that, The safety control method is applied to a target vehicle, which includes a primary driving system and a backup driving system. The safety control method includes: The acquisition module of the target vehicle collects environmental data during the vehicle's driving process and transmits the environmental data to the monitoring module, so that the monitoring module can determine whether the target vehicle is in a warning state based on the environmental data. Calculate the uncertainty parameter σ and dynamic uncertainty threshold σ of the driver system based on the environmental data. th Determine whether the uncertainty parameter σ is greater than the dynamic uncertainty threshold σ. th ; If so, the vehicle in question is determined to be in a warning state; If not, then it is determined that the vehicle is being driven by the main driving system; The environmental data includes traffic density D, vehicle speed change rate V, weather conditions T, and road characteristics W; The uncertainty parameter σ and dynamic uncertainty threshold σ of the master driving system are calculated based on the environmental data. th ,include: A mapping function f is constructed based on the model of the master driving system. DNN The uncertainty σ of the prediction result is obtained through Bayesian deep learning techniques. 2 And follow the formulas 1-2 below: Formula 1: ; Formula 2: ; Where θ represents the deep neural network parameters of the model, X represents the input of the acquisition module, and Y represents the control output; The environmental complexity C is calculated based on the traffic density D, vehicle speed change rate V, weather conditions T, and road characteristics W. The dynamic uncertainty threshold σ is calculated based on the environmental complexity C. th And follow the formula 3 below: Formula 3: ; Where, σ base This serves as a baseline uncertainty threshold, used to replace the dynamic uncertainty threshold σ in simple scenarios. th Use, P human σ represents the probability that the driver will take over the vehicle. total The total uncertainty of multimodal fusion is represented by k1, k2, and α, which are proportionality coefficients. If the vehicle is in a warning state, the driver system is controlled to enter a pre-switching mode to prepare to switch from the driver system to the passenger system. The monitoring module is controlled to monitor the operating status of the target vehicle and determine whether the driver's system has malfunctioned. If so, the control system for the vehicle is switched from the primary driving system to the backup driving system.

2. The safety control method according to claim 1, characterized in that, The master driving system employs a first decision algorithm to handle driving tasks under normal circumstances; The backup driving system employs a second decision algorithm to handle driving tasks in emergency situations; The predictability of the second decision algorithm is higher than that of the first decision algorithm.

3. The safety control method according to claim 1, characterized in that, The acquisition module includes a lidar, a camera, a millimeter-wave radar, and an ultrasonic radar; The data acquisition module that controls the target vehicle collects environmental data during the vehicle's driving process, including: Control the lidar to collect point cloud data; Control the camera to acquire image data; Control the millimeter-wave radar to acquire target detection data; Control the ultrasonic radar to acquire near-field detection data.

4. The safety control method according to claim 3, characterized in that, The data acquisition module that controls the target vehicle collects environmental data during the vehicle's driving process, including: Environmental data is calculated using a Kalman filter algorithm based on the point cloud data, the image data, the target detection data, and the near-field detection data. The weight ratios of the point cloud data, the image data, the target detection data, and the near-field detection data in the Kalman filter algorithm are determined by the vehicle position, speed, and obstacle information of the target vehicle.

5. The safety control method according to claim 4, characterized in that, The step of calculating environmental data using a Kalman filter algorithm based on the point cloud data, the image data, the target detection data, and the near-field detection data includes: Initialize state vector x k The state vector x k The environmental characteristics of the object vehicle are described, and follow the following formula 6: Formula 6: ; Where, p x p y This indicates the vehicle's position on a two-dimensional plane; v x v y Represents the vehicle's velocity component; o xi o yi The coordinates of the i-th obstacle are represented; n is the number of obstacles in the environment; and T represents the transpose of the matrix. The state prediction step is performed using the state transition function f(·), following the formulas 7-8 below: Formula 7: ; Formula 8: ; in, It is the predicted state vector; It is the prediction error covariance matrix; and It is a priori state estimate, represented as a predicted value, without incorporating current measurements; and It is the posterior state estimate, represented as the updated value, which incorporates the current measurement; f(·) refers to the state transition function; It is a control input; It is a Jacobian matrix; This refers to the process noise covariance matrix; k represents the current time, and k-1 represents the previous time. The update step involves calculating the observed value z of the acquisition module s based on the data acquisition module s. s and the corresponding observation matrix H s And follow the formulas 9-13 below: Formula 9: ; Formula 10: ; Formula 11: ; Formula 12: ; Formula 13: ; in, It is the observation noise covariance matrix. It is the Kalman gain, h s (·) is the observation function. It is the measurement noise covariance matrix; Calculate the confidence level w of the acquisition module s. s And follow the formula 14 below: Formula 14: ; in, It is the uncertainty measure of the acquisition module s. It refers to the level of uncertainty. It is a sensitivity parameter used to adjust the steepness of the confidence curve; In the update step, the confidence level w of the acquisition module s is used as a reference. s Kalman gain Perform a weighted average, following the formula 15-17: Formula 15: ; Formula 16: ; Formula 17: ; Where z represents all observations The combination of h(·) is the comprehensive observation function.

6. The safety control method according to claim 1, characterized in that, The target vehicle includes critical modules and non-critical modules, and the target vehicle has a higher priority in the computing power allocation for the critical modules than for the critical modules.

7. An electronic device, characterized in that, The electronic device includes: a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the security control method as described in any one of claims 1 to 6.

8. A storage medium, characterized in that, The storage medium stores a program or instructions that, when executed by a processor, implement the steps of the security control method as described in any one of claims 1 to 6.

Images