Vehicle control method and system considering vehicle and non-vehicle conflict

Abstract

The invention belongs to the field of road vehicle control systems, and relates to a vehicle control method and system considering non-motor vehicle conflict, and the method comprises the steps: firstly obtaining non-motor vehicle driving state information and environment state data, and sensing acute lane changing behaviors, red light running, converse driving and out-of-control danger degrees according to the obtained data; then, based on the driving state information of the non-motor vehicle and the surrounding motor vehicles and the acute lane changing behavior of the non-motor vehicle, the lateral collision danger degree and the front collision danger degree are sensed; various potential conflict scenes between motor vehicles and non-motor vehicles are comprehensively covered, then weights of different risk factors are calculated by using an improved frequency statistical method, comprehensive risk degrees are calculated based on the different risk factors and the weights, and finally, the comprehensive risk degrees are graded and different control strategies are adopted. According to the method, accurate prediction of potential conflicts is realized, and then active intervention is performed according to the prediction result, so that the conflict risk is effectively reduced, and the driving safety is ensured.

Description

Technical Field

[0001] The present invention belongs to the field of road vehicle control systems, and in particular relates to a vehicle control method and system taking into account vehicle-non-machine conflicts. Background Art

[0002] In recent years, with the acceleration of urbanization and the rapid growth of travel demand, the number of non-motor vehicles and motor vehicles has increased sharply, and traffic congestion has become increasingly prominent, which has become a common problem in urban traffic management. As an important means of transportation, non-motor vehicles are favored by short- and medium-distance travelers because of their advantages such as labor-saving, fast, flexible, convenient, efficient, practical, environmentally friendly and economical. According to statistics, the number of electric bicycles in my country has exceeded 400 million, and they occupy an important position in the urban road traffic system at different levels of economic development.

[0003] However, the popularity of non-motorized vehicles has also caused a series of traffic safety problems. Non-motorized vehicles have the characteristics of relatively high speed, strong flexibility, high difficulty in control, and sensitivity to interference. In addition, the cost of their violations is low and the safety protection equipment of riders is insufficient, resulting in frequent non-motorized vehicle accidents and an increasingly severe traffic safety situation. In addition, non-motorized vehicles and motor vehicles often share lanes, leading to chaotic traffic order, which not only reduces traffic efficiency, but also exacerbates the occurrence of collisions between motor vehicles and non-motorized vehicles. The loss of life and economic losses caused by such accidents are far higher than other types of non-motorized vehicle accidents.

[0004] At present, the main methods to solve the conflict between motor vehicles and non-motor vehicles include optimizing traffic planning and design, strengthening management, encouraging public participation, and improving infrastructure. Although these measures can reduce the accident rate and improve road safety and traffic efficiency to a certain extent, they still have limitations, especially when the risk of accidents is high, it is often impossible to take effective measures in advance to avoid them.

[0005] In the prior art, relevant patents are mostly focused on the identification and warning of traffic conditions, but lack the accurate prediction of the risk of collision between non-motor vehicles and motor vehicles and the active control of vehicles, which cannot effectively reduce the risk of accidents. For example, Chinese patent CN 115171383 discloses a "vehicle-road cooperative warning system for dangerous sections", and proposes a vehicle-road cooperative warning system based on millimeter-wave radar and LORA communication modules, which can detect the position, direction and speed of pedestrians, non-motor vehicles and motor vehicles in real time, and transmit data through low-power, anti-interference wireless communication modules; however, non-motor vehicles have strong flexibility, and there are situations such as sudden lane changes and rollovers. Only by detecting the position, direction and speed of pedestrians, non-motor vehicles and motor vehicles, it is not possible to effectively predict the risk of collision between non-motor vehicles and motor vehicles, and it is impossible to actively provide vehicles with effective control strategies to reduce the risk of accidents. Chinese patent CN 116704199 A discloses "a multi-task traffic panoramic perception method, device and computer equipment", which obtains the target driving area, lane line, target location and motion state information through feature extraction and target detection model, determines the traffic panoramic perception information, and improves the perception speed and dimension. Although this method can determine the traffic panoramic perception information, how to accurately predict the collision risk between non-motor vehicles and motor vehicles based on the acquired traffic panoramic perception information has not been disclosed, so it is still impossible to actively provide an effective control strategy for the vehicle; Chinese patent CN 110867081A discloses "a wireless radio frequency identification electric vehicle safety driving device and its working process", which focuses on monitoring the dangerous driving behavior of electric non-motor vehicles and motor vehicles, and transmits warning information through wireless signals, but lacks accurate prediction of the collision risk between non-motor vehicles and motor vehicles.

[0006] Although existing technologies have made some progress in traffic state perception and early warning, they are still insufficient in actively controlling vehicles to avoid accidents because they cannot accurately predict the risk of collision between non-motor vehicles and motor vehicles in mixed lanes. Therefore, it is urgent to develop a vehicle active control method based on accurate and reliable prediction of vehicle-non-motor vehicle conflicts, which has important practical significance for preventing and reducing the occurrence of traffic accidents. Summary of the invention

[0007] In view of the above-mentioned technical problems and defects, the purpose of the present invention is to provide a vehicle control method that takes into account motor vehicle and non-motor vehicle conflicts. This method realizes accurate prediction of potential conflicts by real-time collection and analysis of dynamic behavior data of motor vehicles and non-motor vehicles, combined with intelligent algorithms, and then actively intervenes according to the prediction results, thereby effectively reducing the risk of conflict and ensuring driving safety.

[0008] To achieve the above object, the present invention adopts the following technical solution: A vehicle control method considering non-machine-machine conflicts, the method comprising the following steps: Step 1. Obtain non-motor vehicle driving status information and environmental status data, and perceive dangerous behaviors and danger levels based on the driving status information and environmental status data. The dangerous behaviors include abrupt lane changes. SLC The degree of danger includes the degree of danger of running a red light , the degree of danger of retrograde , the degree of risk of loss of control ; Step 2: Based on the driving status information of non-motor vehicles and surrounding motor vehicles and the abrupt lane-changing behavior of non-motor vehicles SLC , perceive the risk of side collision At the same time, the safety distance D to prevent rear-end collision is calculated based on the driving status information of the non-motor vehicle and the rear motor vehicle, and the degree of front collision risk is perceived based on the safety distance D and the distance between the front non-motor vehicle and the rear motor vehicle. ; Step 3. Calculate the danger level of running a red light based on the probability and severity coefficient of running a red light, driving against traffic, losing control, side collision, and front collision in historical data. , the degree of danger of retrograde , the degree of risk of loss of control , Side collision risk , Frontal collision risk The weight of Step 4. Use a weighted scoring model to add up the weights of each risk factor and the current risk level to obtain a comprehensive risk level score; Step 5. Determine the danger level of the current driving situation based on the comprehensive danger level score. Different danger level intervals correspond to different control strategies.

[0009] As a preferred embodiment of the present invention, a method for sensing the danger level of running a red light is: Step a. Obtaining a video stream collected by a visual module on a non-motor vehicle driver's smart helmet; Step b. Use the YOLO-v8 model to perceive the status information of the traffic lights in the collected video stream; Step c. Calculate the focal length f of the visual module and the imaging height of the traffic light pixel W p The ratio of Step d. Based on the current status of the signal light and Danger level of running red lights for non-motor vehicles Make a judgment: ; in, is the corrected threshold, and the calculation formula is: ; in, v is the speed of non-motor vehicles, α is the empirical coefficient, v max is the maximum driving speed, is the threshold value, which is 0.2.

[0010] As a preferred embodiment of the present invention, the method for sensing the degree of danger of reverse driving is as follows: obtaining the current position data of the non-motor vehicle and the position data of the non-motor vehicle at the next moment through the GPS module, and at the same time obtaining the center line position information of the road on which the non-motor vehicle is currently traveling; sensing the degree of danger of reverse driving of the non-motor vehicle based on the principle of vector product method, and when the non-motor vehicle is detected to be driving in the opposite direction, the degree of danger of reverse driving is detected. ,otherwise .

[0011] As a preferred embodiment of the present invention, when perceiving acute lane change behavior, a static inclination angle threshold or a dynamic lane reference model is selected for judgment based on the lane line. If the lane line is clear and it is a straight road, the static inclination angle threshold model is used for calculation; if the lane is not straight or the lane line cannot be accurately detected, the dynamic lane reference model is used for calculation.

[0012] As a preferred embodiment of the present invention, the perception of the risk of loss of control is determined by the lateral acceleration and turning radius of the non-motor vehicle in combination with the road conditions, and the expression is: ; in, is the radial acceleration of the non-motor vehicle, ; is the current speed of the non-motor vehicle; is the turning radius of non-motor vehicles, is the coefficient of friction of the road surface.

[0013] As a preferred embodiment of the present invention, the method for sensing the risk of lateral collision based on the abrupt lane change behavior of non-motor vehicles is: firstly calculating the lateral distance between the motor vehicle and the adjacent non-motor vehicle; , the lateral distance is corrected based on the abrupt lane change behavior, in, for The corrected lateral distance between the motor vehicle and the adjacent non-motor vehicle, is the distance correction coefficient, which is 0.2. The degree of risk of side collision is determined based on the corrected lateral distance between the motor vehicle and the adjacent non-motor vehicle. The expression is:

[0014] As a preferred embodiment of the present invention, when the front collision risk is perceived, the safe distance to prevent rear-end collision is first determined according to the running state of the non-motor vehicle. Calculate the distance S between the non-motor vehicle in front and the motor vehicle behind and the safety distance Compare and judge the risk of front collision , the expression is: .

[0015] As a preferred embodiment of the present invention, the method for determining the weight coefficient in step 3 is: First, calculate the probability of the jth risk factor appearing in the historical data : ; In the formula, For the j The number of occurrences of risk factors, is the total number of samples; risk factors include running a red light, driving in the wrong direction, loss of control, side collision, and front collision; Then, the weight of each risk factor is calculated, and the weight formula is: ; in, is the weight of the jth risk factor; However, the introduction of the severity factor , the weights are modified, and the final weight formula is: ; in, is the weight of the j-th risk factor after correction.

[0016] As a preferred embodiment of the present invention, in step 5, the comprehensive risk level score is used. The risk is divided into four levels according to the size of the risk. For low risk, the driver is reminded by early warning; for relatively low risk, the deceleration and power output are controlled while the early warning is given; for medium risk, the driver is warned according to the formula Dynamically adjust deceleration, provide steering suggestions, and control power output; for high risks, trigger stepped emergency braking while warning, synchronously activate the electric power steering system, automatically perform obstacle avoidance actions under the constraint of lateral acceleration ≤ 0.3g, and completely cut off power output.

[0017] As a preferred embodiment of the present invention, when the non-motor vehicle ahead is stationary, the safety distance to prevent rear-end collision is The expression is: ； When the non-motor vehicle ahead is traveling at a constant speed or accelerating, the motor vehicle behind is During the time period Reduce the speed to the same as the vehicle in front at all times to avoid rear-end collision. The expression is: ； Otherwise, the safe distance to prevent rear-end collision The expression is: ; When the non-motor vehicle ahead is in a decelerating state, keep a safe distance to prevent rear-end collision The expression is: ; in, and are the speeds of the non-motor vehicle in front and the motor vehicle behind, and are the accelerations of the non-motor vehicle in front and the motor vehicle behind, The reaction time of the motor vehicle driver. is the braking response time, is the braking time when the speed of the non-motor vehicle in front is the same as that of the motor vehicle behind. The minimum safe distance.

[0018] As a further preferred embodiment of the present invention, when the static tilt angle threshold model is calculated, the steering angle and radial acceleration are used for judgment. When the following two conditions are met at the same time, it is considered that an abrupt lane change behavior has occurred: and ; in, is the actual angle between the non-motor vehicle and the lane line, is the radial acceleration of the non-motor vehicle, is the threshold value, which is 0.4g. is the threshold value, which is 20°; The specific steps of dynamic lane benchmark model calculation are as follows: Step a. Dynamically construct a lane benchmark model; First, the lane lines are detected and the road boundaries are extracted using LiDAR point clouds. Then, the detected lane lines or road boundary points are fitted into lane benchmarks using B-spline curve fitting. Cubic uniform B-splines are used for B-spline curve fitting, and the number of control points is dynamically adjusted according to the road curvature. Step b. The lateral displacement change rate of the non-motor vehicle relative to the lane reference Perform calculations; Step c. According to the lane curvature Adaptive adjustment of the maximum allowable deviation rate , the expression is: ; in, is the basic threshold, is the curvature compensation coefficient, is the lane curvature threshold; Step d. Abrupt lane change behavior for non-motor vehicles SLC Make a judgment, if , it is considered that the non-motor vehicle has made an acute lane change. ,otherwise .

[0019] As a further preferred embodiment of the present invention, the control points are optimized by weighted least squares method with Tikhonov regularization, and the objective function is: ; in, is the weighted least squares term, Q represents the coordinate matrix of the road feature points obtained by the sensor, B is the matrix composed of B-spline basis functions, C is the coordinate matrix of the control points to be solved, and the weight matrix W is a diagonal matrix whose diagonal elements are assigned according to the detection confidence of each feature point; is the Tikhonov regularization term. L usually adopts a second-order difference operator matrix to constrain the change range of the control points. The value range of the regularization coefficient λ is set to 0.1 to 0.3.

[0020] The present invention also provides a vehicle control system that considers non-motor vehicle conflicts and is used to implement the control method described above. The system includes an information collection module, a road feature perception module, a vehicle parameter perception module, a non-motor vehicle driving behavior perception module, a danger level perception module, a danger level determination module, and an early warning and control module; wherein the information collection module is used to obtain driving status information and environmental status data of motor vehicles and non-motor vehicles; The road feature perception module is used to obtain road related information, including road center line, lane line, lane curvature, and road boundary points, based on the environmental status data collected by the information collection module; The vehicle parameter perception module is used to obtain various dynamic parameters of the interaction between the motor vehicle and the non-motor vehicle during the driving process, including driving speed, acceleration, and steering angle, based on the driving state information of the motor vehicle and the non-motor vehicle collected by the information collection module; The non-motor vehicle driving behavior perception module is used to perceive the dangerous behavior and degree of danger during the driving of the non-motor vehicle according to the non-motor vehicle driving state data and the environmental state data; The danger level perception module is used to determine the danger level of the front collision and the danger level of the lateral collision according to the dynamic parameters obtained by the vehicle parameter perception module and the dangerous behavior perceived by the non-motor vehicle driving behavior perception module; The danger level determination module is used to calculate the comprehensive danger level according to the danger levels of front collision, side collision, loss of control, running a red light and driving against traffic, and classify the danger level of the current traffic conditions in combination with the preset threshold value; The early warning and control module is used to take corresponding early warning and control measures according to the result of the danger level determination module.

[0021] Advantages and beneficial effects of the present invention: (1) The present invention realizes accurate prediction of potential conflicts and improves the reliability of detection results by real-time collection and analysis of dynamic behavior data of motor vehicles and non-motor vehicles and combining intelligent algorithms.

[0022] (2) In order to break through the dependence of existing technologies on scenarios, the present invention avoids using a single data source or parameters that are easily affected by the environment. Instead, it integrates multi-dimensional data to construct a universal conflict detection model suitable for complex urban road environments, ensuring the stability and adaptability of the system in different scenarios.

[0023] (3) The present invention establishes a multi-type conflict model, which not only covers behaviors such as running red lights in traditional scenarios, but also further expands to more complex and dangerous behaviors such as driving in the wrong direction and changing lanes, comprehensively covering all types of potential conflict scenarios between motor vehicles and non-motor vehicles, thereby significantly improving the system's prediction accuracy of potential conflicts.

[0024] (4) To solve the problem that the existing technology only provides warning but lacks active intervention, the present invention combines intelligent transportation and vehicle-mounted communication technologies to send warning information to the driver and dynamically coordinate the driving status of the motor vehicle according to the predicted risk level, conduct active intervention, effectively reduce the risk of conflict, and ensure driving safety.

[0025] (5) The present invention detects the abrupt lane change behavior and predicts the risk level of the lateral collision based on the abrupt lane change behavior, thereby making the prediction of the risk level of the lateral collision more accurate.

[0026] (6) The present invention calculates the safe distance to prevent rear-end collision based on the running status of the non-motor vehicle in front, and predicts the risk of front collision according to the safe distance and the distance between the non-motor vehicle in front and the motor vehicle behind. This method can produce accurate and reliable prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By referring to the following description in conjunction with the accompanying drawings, and with a more comprehensive understanding of the present invention, other objects and results of the present invention will become more apparent and easy to understand. In the accompanying drawings: Figure 1 A flow chart of a vehicle control method considering machine-non-conflict provided by the present invention; Figure 2 A structural block diagram of a vehicle control system considering machine-non-conflict provided by the present invention. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the technical solution and advantages of the present invention, the present application is described in detail below in conjunction with the accompanying drawings, but it is not intended to limit the protection scope of the present invention.

[0029] Embodiment 1: like Figure 1 As shown, this embodiment provides a vehicle control method considering non-machine-machine conflicts, and the method includes the following steps: Step 1. Obtain non-motor vehicle driving status information and environmental status data, and perceive dangerous behaviors and danger levels based on the driving status information and environmental status data. The dangerous behaviors include abrupt lane changes. SLC The degree of danger includes the degree of danger of running a red light , the degree of danger of retrograde , the degree of risk of loss of control ; Step 2: Based on the driving status information of non-motor vehicles and surrounding motor vehicles and the abrupt lane-changing behavior of non-motor vehicles SLC , perceive the risk of side collision At the same time, the safety distance D to prevent rear-end collision is calculated based on the driving status information of the non-motor vehicle and the rear motor vehicle, and the degree of front collision risk is perceived based on the safety distance D and the distance between the front non-motor vehicle and the rear motor vehicle. ; Step 3. Calculate the danger level of running a red light based on the probability and severity coefficient of running a red light, driving against traffic, losing control, side collision, and front collision in historical data. , the degree of danger of retrograde , the degree of risk of loss of control , Side collision risk , Frontal collision risk The weight of Step 4. Use a weighted scoring model to add up the weights of each risk factor and the current risk level to obtain a comprehensive risk level score; Step 5. Determine the danger level of the current driving situation based on the comprehensive danger level score. Different danger level intervals correspond to different control strategies.

[0030] In this embodiment, the vision module (camera) and the GPS module are installed on the smart helmet of the non-motor vehicle driver. The driver's smart helmet is linked with the on-board networking facilities to obtain the driving status information of the non-motor vehicle during driving through the on-board networking facilities, and perceive dangerous behaviors and risk levels based on the driving status information.

[0031] In this embodiment, the process of sensing the danger level of running a red light during driving of a non-motor vehicle is as follows: Step a. Obtain the video stream (resolution) collected by the visual module (camera) on the non-motor vehicle driver's smart helmet H × W ,focal length f ); Step b. Use the YOLO-v8 model to perceive the status information of the traffic lights in the collected video stream; Specifically, by modifying the format of the output header of the YOLO-v8 model, the YOLO-v8 model synchronously generates the following information:

[0032] in: (x,y) is the coordinate of the center point of the traffic light, (w,h) The width and height of the traffic light bounding box, color prob is the probability of a red light (Sigmoid output); y top , y bottom for The top and bottom ordinates of the traffic light bounding box.

[0033] The traffic light status is monitored through the generated information. S Make a judgment:

[0034] in, S ∈{0 (green light), 1 (red light)}; Step c. Based on the principle of monocular ranging, the focal length f of the visual module (camera) and the imaging height of the traffic light pixel are calculated. W p The ratio k reflects the relative distance between non-motor vehicles and traffic lights; because:

[0035] in, W p is the pixel imaging height of the traffic light, extracted through the target detection model, W r The physical height of the traffic light ； As the distance d decreases, Continuously increasing, Keep decreasing, when it decreases to the threshold value determined by the experiment When the vehicle is driving in a red light, it can be considered that a red light running behavior has occurred.

[0036] Since there are situations where high-speed non-motor vehicles cannot brake in time, it is necessary to adjust the threshold Make corrections so that it can provide early warning and control of other motor vehicles:

[0037] in, v is the speed of non-motor vehicles, α is the empirical coefficient, and its value is 0.2; v max is the maximum travel speed, which is 15 m / s; The value is 0.2.

[0038] Step d. The degree of danger of non-motor vehicles running red lights based on the current state of the traffic light and the ratio k Make a judgment:

[0039] in, is the corrected threshold.

[0040] In this embodiment, when a non-motor vehicle is detected to have run a red light, the danger level of running a red light is ,otherwise .

[0041] In this embodiment, the process of sensing the danger level of reverse driving (reverse driving) during non-motor vehicle driving is as follows: In the perception of the danger level of reverse driving, the current location data of non-motor vehicles is obtained through the GPS module The location data of non-motor vehicles at the next moment At the same time, the centerline position information of the road where the non-motor vehicle is currently traveling is obtained through the vehicle networking facilities ; Based on the principle of vector product method, the danger level of non-motor vehicles driving in the opposite direction is perceived. The specific modeling method is as follows: Step a. Coordinate system conversion (WGS84 to plane coordinates): The WGS84 coordinate system is a global three-dimensional coordinate system that is widely used in geographic positioning, surveying and mapping standards, and the Global Positioning System (GPS). The present invention converts WGS84 longitude and latitude into plane coordinates, and can convert the coordinates returned by the map API (map application program interface) into plane coordinates used for calculating vector cross products, thereby avoiding errors caused by the curvature of the earth.

[0042] Specifically, the present invention uses the transverse Mercator projection formula, and the core formula is:

[0043]

[0044] in: is the latitude and longitude, is the UTM scale factor, which is 0.9996. The UTM (Universal Transverse Mercator) projection is a conformal cylindrical projection that projects the earth's surface onto a cylinder and then unfolds it into a plane. is the meridian arc length integral formula, is the first eccentricity, is the semi-major axis of the Earth ellipsoid, is the semi-minor axis of the Earth ellipsoid, is the radius of curvature of the Maoyou circle, is the second eccentricity of the Earth, ( x,y ) is the coordinate of the plane coordinate system after transformation; Step b. Projection method and closest point algorithm: The projection method can optimize the search for the coordinates of the vertical point from the driver's position to the center line of the road at time t, and enhance the judgment accuracy in conditions such as curved roads. The specific formula is as follows: Given a line segment endpoint and non-motorized vehicle coordinates :

[0045]

[0046] in, represents the proportional position of the foot of the perpendicular on line segment AB, Represents the center line vertical point Non-motor vehicle location The direction vector, Represents the center line vertical point With endpoint The direction vector, Represents the center line vertical point With endpoint The direction vector of Step c. Determine the degree of retrograde danger by using the two-dimensional vector cross multiplication method :

[0047] in, For the same, is the displacement vector of non-motor vehicles during this period.

[0048] In this embodiment, when a non-motor vehicle is detected to be traveling in the wrong direction, the degree of danger of traveling in the wrong direction is ,otherwise .

[0049] In this embodiment, the process of sensing the abrupt lane change behavior during the driving of a non-motor vehicle is as follows: In order to reduce the amount of calculation and to provide early warning and active intervention control for motor vehicles more quickly, the present invention detects sudden lane changes (SLC) by using a static inclination angle threshold or a dynamic lane reference model according to the lane line. If the lane line is clear and it is a straight road, the static inclination angle threshold model is used for calculation; if the lane is not straight or the lane line cannot be accurately detected, the dynamic lane reference model is used for calculation. The specific situation is as follows: (1) Static tilt threshold model calculation: The model is mainly judged by two variables: steering angle and radial acceleration.

[0050] When the driver changes lanes suddenly, the steering angle α and the angle β between the non-motor vehicle and the lane line are complementary in value:

[0051] in, is the steering angle of the vehicle; is the angle between the actual non-motor vehicle and the lane line; Due to perspective, the angle calculated in the image Not equal to the actual angle , needs to be corrected by top view transformation:

[0052] Among them, Transform represents the transformation process of converting the lane lines in the image into a top-down view and then calculating the actual angle.

[0053] when Less than the experimentally determined threshold When the vehicle's steering angle is too large, the IMU module is used to detect the radial acceleration of non-motor vehicles. ; When the radial acceleration detected Greater than the threshold set by the experiment When , it is considered that the vehicle has a large radial acceleration. An abrupt lane change is considered to have occurred when both of the following conditions are met: and

[0054] in, The value is 0.4g, The value is 20°; (2) Dynamic lane benchmark model calculation: First, the lane benchmark model is dynamically constructed; Specifically, the lane lines are detected using camera visual perception, and the road boundaries are extracted using LiDAR point clouds. Subsequently, the detected lane lines or road boundary points are fitted into a virtual benchmark (lane benchmark) using B-spline curve fitting, and its function is expressed as:

[0055] in, is the B-spline basis function, is the control point, is a parameter, n is the number of collection points, For a B-spline curve with parameters The coordinate value at the location; in the specific implementation process of the dynamic lane benchmark model, the system first constructs road features through multi-sensor data fusion; specifically, the camera collects images at a frame rate of 60Hz, and uses a deep learning-based segmentation network to extract lane feature points. At the same time, the 64-line LiDAR obtains the three-dimensional coordinates of the road boundary through point cloud clustering; the two types of data are uniformly processed in the vehicle coordinate system after spatiotemporal registration.

[0056] In this embodiment, the B-spline curve fitting uses cubic uniform B-spline, and the number of control points is dynamically adjusted according to the road curvature. The control points are optimized by weighted least squares method with Tikhonov regularization, and the objective function is:

[0057] in, is the weighted least squares term, Q represents the coordinate matrix of the road feature points obtained by the sensor, B is the matrix composed of B-spline basis functions, C is the coordinate matrix of the control points to be solved, and the weight matrix W is a diagonal matrix whose diagonal elements are assigned according to the detection confidence of each feature point.

[0058] is the Tikhonov regularization term. L usually adopts a second-order difference operator matrix to constrain the change range of the control points. The value range of the regularization coefficient λ is set to 0.1 to 0.3. By adjusting this parameter, the relationship between curve fitting accuracy and smoothness can be balanced.

[0059] Then, Cholesky decomposition is used to efficiently solve the problem. The normal equation corresponding to the objective function is: .

[0060] Then, based on the calculation results, the lateral displacement change rate of the vehicle (non-motor vehicle) relative to the lane reference (virtual reference) is calculated. Calculation is performed to reflect the urgency of lane change for non-motor vehicles:

[0061] in, Unit time The lateral distance change between the vehicle and the lane benchmark.

[0062] Then, according to the lane curvature Adaptive adjustment of the maximum allowable deviation rate , the expression is:

[0063] in, is the basic threshold, the value is 0.7m / s, is the curvature compensation coefficient, the value is 0.2, is the lane curvature threshold, with a value of 0.03 .

[0064] Finally, the sudden lane change behavior of non-motor vehicles SLC Make a judgment:

[0065] In this embodiment, if , it is considered that the non-motor vehicle has made an acute lane change. ,otherwise .

[0066] In this embodiment, the risk level of lateral collision is sensed based on the abrupt lane change behavior of non-motor vehicles, and the specific situation is as follows: For the risk of side collision, the present invention considers the distance between the motor vehicle and the adjacent non-motor vehicle and the driving behavior of the non-motor vehicle, especially the danger of changing lanes. The risk of side collision is evaluated by measuring the lateral distance and judging whether an abrupt lane change occurs.

[0067] The lateral distance between a motor vehicle and an adjacent non-motor vehicle is calculated as follows:

[0068] In the formula, is the lateral distance between the motor vehicle and the adjacent non-motor vehicle (m); , is the horizontal and vertical coordinates of the current motor vehicle; , are the horizontal and vertical coordinates of the adjacent non-motor vehicles.

[0069] At the same time, if a non-motor vehicle changes lanes suddenly, the risk of side collision will increase, so the lateral distance needs to be corrected:

[0070] Among them, SLC is the marker variable of acute lane change behavior, is the distance correction coefficient, which is taken as 0.2 here.

[0071] Side collision risk The evaluation looks like this:

[0072] In this embodiment, when there is a risk of side collision, =1, otherwise, =0.

[0073] In this embodiment, the process of sensing the degree of front collision risk during the driving of a non-motor vehicle is as follows: In view of the degree of front collision risk, the present invention uses ultrasonic sensors to sense the speed and relative speed of non-motor vehicle drivers, constructs a safety distance model, determines the values ​​of various parameters in the model, and adopts different safety distances for different situations; the safety distance calculation formula of the rear-end collision risk perception module is:

[0074] in, To prevent rear-end collisions, and are the distances traveled by the front vehicle (non-motor vehicle) and the rear vehicle (motor vehicle) during the braking process, It is the minimum safety distance, and its value is 2m.

[0075] The car is currently stationary:

[0076]

[0077] in, is the speed of the front vehicle (non-motor vehicle), is the acceleration of the front vehicle (non-motor vehicle), The reaction time of the motor vehicle driver. is the brake response time; The current vehicle is in a state of constant speed or constant acceleration: The relative speed of the two vehicles is relatively small, and the rear vehicle is A moment in a time period The speed of the vehicle ahead is reduced to the same speed as the vehicle ahead. During the time period, the deceleration of the following vehicle is directly proportional to time;

[0078] in, and are the speeds of the front and rear vehicles, respectively. The reaction time of the motor vehicle driver. is the braking response time, is the braking time when the speed of the rear vehicle is the same as that of the front vehicle; The relative speed of the vehicles is relatively high. After the deceleration of the rear vehicle reaches the maximum value, its speed is still higher than that of the front vehicle. At this time, the maximum braking deceleration needs to be maintained for a period of time before the rear vehicle can slow down to the same speed as the front vehicle.

[0079]

[0080] The current car is in a decelerating motion state: Since the front car is in a decelerating state, as long as either of the two cars is still moving, the distance between the two cars will continue to shrink. Therefore, the moment when the speed of the rear car and the front car is reduced to zero (both stop) is the most dangerous moment.

[0081]

[0082] in, and are the speeds of the front and rear vehicles, respectively. and are the accelerations of the front and rear vehicles, respectively. The reaction time of the motor vehicle driver. is the brake response time.

[0083] Frontal collision risk The evaluation can be determined by the safety distance D to prevent rear-end collision, as shown below.

[0084] in, S It is the distance between the non-motor vehicle in front and the motor vehicle behind.

[0085] In this embodiment, the process of sensing the risk of loss of control during the driving of a non-motor vehicle is as follows: The risk of loss of control is usually related to factors such as the speed, turning angle, and road friction coefficient of the non-motor vehicle. The present invention predicts whether there is a possibility of loss of control by judging the lateral acceleration and turning radius of the non-motor vehicle and combining it with the road conditions. The expression is:

[0086] in, is the radial acceleration of the non-motor vehicle (m / s 2 ), ; is the current speed of the non-motor vehicle (m / s); is the turning radius of non-motor vehicles (m), is the friction coefficient of the road surface, In normal weather conditions, take 0.7; in rainy days, take 0.3; and in snowy days, take 0.1.

[0087] In order to comprehensively assess the risk , using a weighted scoring model, the weights of each risk factor and the current risk level are weighted and summed to obtain a comprehensive risk level score. However, due to the risk level vector It is a binary variable of 0-1. The common entropy weight method may fail to calculate each risk factor. Therefore, the present invention uses an improved frequency statistics method to calculate the weight of each risk factor. The specific situation is: First, calculate the probability of the jth risk factor appearing in the historical data :

[0088] In the formula, For the j The number of occurrences of risk factors, is the total sample number; Then, the weight of each risk factor is calculated. The higher the frequency of the risk, the lower its weight should be; conversely, rare but high-risk events should be given a higher weight. Therefore, the weight formula is:

[0089] in, is the weight of the jth risk factor; However, considering that some risks occur infrequently but have serious consequences, the weights need to be modified in combination with expert knowledge, so the severity coefficient is introduced. , the final weight formula is:

[0090] in, is the weight of the j-th risk factor after correction; Finally, the comprehensive risk score It can be expressed as: ; in, , , , , They are the weights of the revised risk factors of running a red light, driving in the wrong direction, rear-end collision, side collision and loss of control.

[0091] In this embodiment, the danger level of the current driving situation is determined based on the size of the comprehensive danger level score, and the specific situation is as follows: A: ; B: ; C: ; D: .

[0092] In this embodiment, different danger level intervals correspond to different control strategies, specifically: (1) Low risk - D range: Provide minimum warning: a short whistle and the corresponding warning icon on the instrument panel flash slightly three times; the accelerator pedal damping is increased by 20% as a tactile reminder, but it does not affect the actual power output.

[0093] (2) Lower risk - Zone C: Preventive measures are mainly taken: a comfortable braking strategy is adopted, and the deceleration is controlled within the range of 0.5~1m / s²; a yellow floating warning box is displayed on the instrument panel display, which can mark the estimated collision time; the accelerator pedal will apply pulsed vibration feedback (each lasting 200ms), and the power output will gradually drop to 50%; if it is detected that the driver has braked more than 3 times within 3 minutes, the intervention intensity will be automatically reduced by 30% to avoid interfering with the driving rhythm.

[0094] (3) Medium risk - Zone B: Implementing progressive defense: Braking system by formula Dynamically adjust deceleration; the steering system provides 5°~10° steering suggestions based on TTC (time to collision), and works with ESP (electronic stability program) to maintain stability; power output is limited to 30% opening, the orange warning area of ​​the instrument panel display screen continues to flash, accompanied by a voice prompt of "Danger! Please brake"; when it is detected that the driver continues to step on the accelerator, the steering wheel will produce 5Hz damping feedback.

[0095] (4) High risk - Zone A: The highest level of intervention is initiated immediately: first, a stepped emergency brake is triggered, with a slow brake of 2m / s² applied for the first 0.5 seconds to avoid tailspin, and then the deceleration is based on the formula The deceleration rate is dynamically increased according to the real-time distance; the electric power steering system is activated synchronously, and the obstacle avoidance action is automatically performed under the constraint of lateral acceleration ≤ 0.3g; the power system completely cuts off the power output and starts the multi-modal alarm at the same time; the front windshield projects a full-screen red countdown, the seat vibrates at a frequency of 10Hz, and emits a high-frequency buzzing sound of 2000Hz. The response delay of the system is strictly controlled within 100ms, ensuring that the vehicle speed is reduced by more than 60% within 1 second.

[0096] The present invention utilizes external precise timing interruption to realize the timed data collection and storage of smart helmets, and transmits the dangerous behavior and danger level data of non-motor vehicles to nearby motor vehicles through the V2X (Vehicle-to-Everything) technology of the Internet of Vehicles, so as to facilitate motor vehicles to analyze the danger level and control the motor vehicles according to the analysis results to avoid collision risks.

[0097] Embodiment 2: like Figure 2 As shown, the present invention also provides a vehicle control system considering the conflict between motor vehicles and non-motor vehicles, the system includes an information collection module, a road feature perception module, a vehicle parameter perception module, a non-motor vehicle driving behavior perception module, a danger level perception module, a danger level determination module, and an early warning and control module; wherein the information collection module is used to obtain the driving state information and environmental state data of motor vehicles and non-motor vehicles; The road feature perception module is used to obtain road related information, including road center line, lane line, lane curvature, road boundary points, etc., based on the environmental status data collected by the information collection module; The vehicle parameter perception module is used to obtain various dynamic parameters of the interaction between the motor vehicle and the non-motor vehicle during the driving process, including parameters such as driving speed, acceleration, and steering angle, based on the driving state information of the motor vehicle and the non-motor vehicle collected by the information collection module; The non-motor vehicle driving behavior perception module is used to perceive the dangerous behavior and degree of danger during the driving of the non-motor vehicle according to the non-motor vehicle driving state data and the environmental state data; The danger level perception module is used to determine the danger level of the front collision and the danger level of the lateral collision according to the dynamic parameters obtained by the vehicle parameter perception module and the dangerous behavior perceived by the non-motor vehicle driving behavior perception module; The danger level determination module is used to calculate the comprehensive danger level according to the danger levels of front collision, side collision, loss of control, running a red light and driving against traffic, and classify the danger level of the current traffic conditions in combination with the preset safety standards and thresholds; The early warning and control module is used to take corresponding early warning and control measures according to the result of the danger level determination module.

[0098] Furthermore, in this embodiment, the information collection module includes a visual module (camera) installed on the non-motor vehicle driver's smart helmet, a GPS module, and vehicle-mounted sensors installed on the non-motor vehicle, the vehicle-mounted sensors including an IMU module, etc.; it also includes vehicle-mounted sensors installed on motor vehicles, etc.

[0099] In this embodiment, the system also includes a memory, and the data obtained by the road feature perception module and the vehicle parameter perception module are stored in the corresponding memory one; the determination result of the danger level determination module is stored in the memory two for use by the subsequent warning and control module.

[0100] The present invention also provides an electronic device, comprising: one or more processors and a memory; wherein the memory is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned vehicle control method.

[0101] The present invention also provides a computer-readable medium having a computer program stored thereon, wherein the computer program implements the above-mentioned vehicle control method when executed by a processor.

[0102] Those skilled in the art will appreciate that all or part of the functions of the various methods / modules in the above embodiments may be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program may be stored in a computer-readable storage medium, which may include: a read-only memory, a random access memory, a disk, an optical disk, a hard disk, etc. The program is executed by a computer to implement the above functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above functions can be implemented.

[0103] In addition, when all or part of the functions in the above-mentioned embodiments are implemented by means of a computer program, the program can also be stored in a storage medium such as a server, another computer, a disk, an optical disk, a flash drive or a mobile hard disk, and saved to the memory of a local device by downloading or copying, or the system of the local device is updated. When the program in the memory is executed by the processor, all or part of the functions in the above-mentioned embodiments can be implemented.

[0104] The above specific examples are used to illustrate the present invention, which is only used to help understand the present invention and is not intended to limit the present invention. For those skilled in the art of the present invention, according to the idea of ​​the present invention, some simple deductions, deformations or substitutions can be made. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A vehicle control method considering machine-non-machine conflict, characterized in that: The method comprises the following steps: Step 1. Obtain non-motor vehicle driving status information and environmental status data, and perceive dangerous behaviors and danger levels based on the driving status information and environmental status data. The dangerous behaviors include abrupt lane changes. SLC The degree of danger includes the degree of danger of running a red light , the degree of danger of retrograde , the degree of risk of loss of control ; Step 2: Based on the driving status information of non-motor vehicles and surrounding motor vehicles and the abrupt lane-changing behavior of non-motor vehicles SLC , perceive the risk of side collision At the same time, the safety distance D to prevent rear-end collision is calculated based on the driving status information of the non-motor vehicle and the rear motor vehicle, and the degree of front collision risk is perceived based on the safety distance D and the distance between the front non-motor vehicle and the rear motor vehicle. ; Step 3. Calculate the danger level of running a red light based on the probability and severity coefficient of running a red light, driving against traffic, losing control, side collision, and front collision in historical data. , the degree of danger of retrograde , the degree of risk of loss of control , Side collision risk , Frontal collision risk The weight of Step 4. Use a weighted scoring model to add up the weights of each risk factor and the current risk level to obtain a comprehensive risk level score; Step 5. Determine the danger level of the current driving situation based on the comprehensive danger level score. Different danger level intervals correspond to different control strategies.

2. A vehicle control method considering machine-non-conflict according to claim 1, characterized in that: The method of perceiving the danger level of running a red light is as follows: Step a. Obtaining a video stream collected by a visual module on a non-motor vehicle driver's smart helmet; Step b. Use the YOLO-v8 model to perceive the status information of the traffic lights in the collected video stream; Step c. Calculate the focal length f of the visual module and the imaging height of the traffic light pixel W p The ratio of Step d. Based on the current status of the signal light and Danger level of running red lights for non-motor vehicles Make a judgment: ； in, is the corrected threshold, and the calculation formula is: ； in, v is the speed of non-motor vehicles, α is the empirical coefficient, v max is the maximum driving speed, is the threshold value, which is 0.

2.

3. The vehicle control method considering non-machine conflicts according to claim 1, characterized in that: The method for sensing the degree of danger of reverse driving is as follows: the current and next moment location data of the non-motor vehicle are obtained through the GPS module, and the center line location information of the road the non-motor vehicle is currently traveling on is obtained at the same time; the degree of danger of reverse driving of the non-motor vehicle is sensed based on the principle of vector product method, and when the non-motor vehicle is detected to be driving in reverse, the degree of danger of reverse driving is ,otherwise ; When sensing a sudden lane change, the driver will be judged by the static inclination angle threshold or the dynamic lane reference model based on the lane line. If the lane line is clear and it is a straight road, the static inclination angle threshold model will be used for calculation. If the lane is not straight or the lane line cannot be accurately detected, the dynamic lane reference model is used for calculation; The perception of the risk of loss of control is determined by the lateral acceleration and turning radius of the non-motor vehicle combined with the road conditions. The expression is: ； in, is the radial acceleration of the non-motor vehicle, ; is the current speed of the non-motor vehicle; is the turning radius of non-motor vehicles, is the coefficient of friction of the road surface.

4. The vehicle control method considering machine-non-conflict according to claim 1, characterized in that: The method for perceiving the risk of lateral collision based on the abrupt lane change behavior of non-motor vehicles is as follows: first, the lateral distance between the motor vehicle and the adjacent non-motor vehicle is calculated. , the lateral distance is corrected based on the abrupt lane change behavior, in, is the corrected lateral distance between the motor vehicle and the adjacent non-motor vehicle, is the distance correction coefficient, which is set to 0.

2. The lateral distance between the motor vehicle and the adjacent non-motor vehicle is used to determine the risk of a side collision. , the expression is: ; When sensing the risk of a front collision, the operator first determines the safe distance to prevent rear-end collision based on the operating status of the non-motor vehicle. Calculate the distance S between the non-motor vehicle in front and the motor vehicle behind and the safety distance Compare and judge the risk of front collision , the expression is: .

5. The vehicle control method considering machine-non-conflict according to claim 1, characterized in that: The weights in step 3 are determined as follows: First, calculate the probability of the jth risk factor appearing in the historical data : ； In the formula, For the j The number of times a risk factor occurs, is the total number of samples; risk factors include running a red light, driving in the wrong direction, loss of control, side collision, and front collision; Then, the weight of each risk factor is calculated, and the weight formula is: ； in, is the weight of the jth risk factor; Then, the severity coefficient is introduced , the weights are modified, and the final weight formula is: ； in, is the weight of the j-th risk factor after correction.

6. The vehicle control method considering non-machine conflicts according to claim 1, characterized in that: In step 5, the comprehensive risk score is used The risk is divided into four levels according to the size of the risk. For low risk, the driver is reminded by early warning; for relatively low risk, the deceleration and power output are controlled while the early warning is given; for medium risk, the driver is warned according to the formula Dynamically adjust deceleration, provide steering suggestions, and control power output; for high risks, trigger stepped emergency braking while warning, synchronously activate the electric power steering system, automatically perform obstacle avoidance actions under the constraint of lateral acceleration ≤ 0.3g, and completely cut off power output.

7. The vehicle control method considering non-machine conflicts according to claim 1, characterized in that: When the non-motor vehicle in front is stationary, keep a safe distance to prevent rear-end collision The expression is: ； When the non-motor vehicle ahead is traveling at a constant speed or accelerating, the motor vehicle behind is During the time period Reduce the speed to the same as the vehicle in front at all times to avoid rear-end collision. The expression is: ; Otherwise, the safe distance to prevent rear-end collision The expression is: ； When the non-motor vehicle ahead is in a decelerating state, keep a safe distance to prevent rear-end collision The expression is: ； in ， and are the speeds of the non-motor vehicle in front and the motor vehicle behind, and are the accelerations of the non-motor vehicle in front and the motor vehicle behind, The reaction time of the motor vehicle driver. is the braking response time, is the braking time when the speed of the non-motor vehicle in front is the same as that of the motor vehicle behind. The minimum safe distance.

8. The vehicle control method considering non-machine conflicts according to claim 3, characterized in that: When the static tilt angle threshold model is calculated, the steering angle and radial acceleration are used for judgment. When the following two conditions are met at the same time, it is considered that an abrupt lane change has occurred: and ; in, is the actual angle between the non-motor vehicle and the lane line, is the radial acceleration of the non-motor vehicle, is the threshold value, which is 0.4g. is the threshold value, which is 20°; The specific steps of dynamic lane benchmark model calculation are as follows: Step a. Dynamically construct a lane benchmark model; First, the lane lines are detected and the road boundaries are extracted using LiDAR point clouds. Then, the detected lane lines or road boundary points are fitted into lane benchmarks using B-spline curve fitting. Cubic uniform B-splines are used for B-spline curve fitting, and the number of control points is dynamically adjusted according to the road curvature. Step b. The lateral displacement change rate of the non-motor vehicle relative to the lane reference Perform calculations; Step c. According to the lane curvature Adaptive adjustment of the maximum allowable deviation rate , the expression is: ； in, is the basic threshold, is the curvature compensation coefficient, is the lane curvature threshold; Step d. Abrupt lane change behavior for non-motor vehicles SLC Make a judgment, if , it is considered that the non-motor vehicle has made an acute lane change. ,otherwise .

9. The vehicle control method considering non-machine conflicts according to claim 8, characterized in that: The control points are optimized by weighted least squares method with Tikhonov regularization, and the objective function is: ； in, is the weighted least squares term, Q represents the coordinate matrix of the road feature points obtained by the sensor, B is the matrix composed of B-spline basis functions, C is the coordinate matrix of the control points to be solved, and the weight matrix W is a diagonal matrix whose diagonal elements are assigned according to the detection confidence of each feature point; is the Tikhonov regularization term, L uses a second-order difference operator matrix to constrain the change range of the control points; the value range of the regularization coefficient λ is set to 0.1 to 0.

3.

10. A vehicle control system considering machine non-conflict for implementing the control method described in any one of claims 1 to 9, characterized in that: The system includes an information collection module, a road feature perception module, a vehicle parameter perception module, a non-motor vehicle driving behavior perception module, a danger level perception module, a danger level determination module, and an early warning and control module; wherein the information collection module is used to obtain driving status information of motor vehicles and non-motor vehicles and environmental status data; The road feature perception module is used to obtain road related information, including road center line, lane line, lane curvature, and road boundary points, based on the environmental status data collected by the information collection module; The vehicle parameter perception module is used to obtain various dynamic parameters of the interaction between the motor vehicle and the non-motor vehicle during the driving process, including driving speed, acceleration, and steering angle, based on the driving state information of the motor vehicle and the non-motor vehicle collected by the information collection module; The non-motor vehicle driving behavior perception module is used to perceive the dangerous behavior and degree of danger during the driving of the non-motor vehicle according to the non-motor vehicle driving state data and the environmental state data; The danger level perception module is used to determine the danger level of the front collision and the danger level of the lateral collision according to the dynamic parameters obtained by the vehicle parameter perception module and the dangerous behavior perceived by the non-motor vehicle driving behavior perception module; The danger level determination module is used to calculate the comprehensive danger level according to the danger levels of front collision, side collision, loss of control, running a red light and driving against traffic, and classify the danger level of the current traffic conditions in combination with the preset threshold value; The early warning and control module is used to take corresponding early warning and control measures according to the result of the danger level determination module.

Images