Crossroad anti-collision early warning method based on trajectory prediction
By acquiring and analyzing the trajectory data of all traffic participants at the intersection in real time, using LSTM and GRU models to predict conflict points and arrival time, the problem of poor reliability in traditional methods is solved, and more accurate early warning and high driving safety is achieved.
Patent Information
- Application Number
- CN202411073195.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-06
- Publication Date
- 2024-12-06
AI Technical Summary
The traditional crossroad collision prevention warning method based on trajectory prediction failed to consider the severity of conflict and comprehensive risks, resulting in poor reliability of early warning strategies.
The data acquisition module obtains the image data of all traffic participants' driving trajectories at the intersection in real time, uses LSTM and GRU for encoding and decoding, analyzes and generates prediction trajectories, calculates the trajectory overlap point and collision arrival time, outputs early warning signals, considering the driver's characteristics, Relationships with traffic participants.
It improves the accuracy of trajectory conflict prediction and the reliability of early warning, and provides two-level early warnings to improve driving safety.
Smart Images

Figure CN119091682A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of traffic management, and in particular to an intersection anti-collision warning method based on trajectory prediction. Background Art
[0002] In the process of traffic management, intersections are usually high-incidence scenes of safety accidents. Therefore, formulating a more reasonable intersection collision warning strategy CICW is of great significance to improving traffic safety at intersections. The mainstream formulation methods include but are not limited to detection methods based on time proximity and inspection methods based on spatial proximity, such as predicting control parameters, deriving collision probability, or analyzing collision risks based on physical field strength. The intersection collision warning strategy CICW mainly includes four aspects: conflict detection, collision warning, test evaluation, and reliability. First, the recurrent neural network long short-term memory neural network LSTM and the gated recurrent unit GRU are used to encode and decode the existing driving trajectory characteristics of traffic participants to regenerate the predicted trajectory. Conflict detection can predict the collision risk more accurately based on the relationship between vehicle status and traffic risk. Collision warning focuses on how to reasonably design warning strategies based on the detected collision risk to provide timely and accurate warnings to drivers. The test evaluation mainly evaluates the effectiveness of CICW from two aspects: collision detection results and the degree of conflict improvement. The reliability of CICW is easily affected by driver behavior and communication uncertainty, which reduces reliability. Driver behavior is difficult to predict, and different types of drivers will react differently when faced with danger and warnings.
[0003] At present, the traditional intersection collision avoidance warning method based on trajectory prediction fails to consider the severity of the conflict and the comprehensive risk. The existing driving safety data all comes from the driving line of sight of a single vehicle body. The data source is single, and a single vehicle body cannot obtain more comprehensive driving safety information other than its own vehicle. In the actual warning process, the warning method does not take into account the characteristics of the driver, and the reliability of the warning strategy is poor. Summary of the invention
[0004] 1. Technical issues to be resolved
[0005] In view of the shortcomings of the prior art, the present invention provides an intersection collision avoidance warning method based on trajectory prediction, which has the advantages of high prediction trajectory conflict accuracy and high two-level warning driving safety. It solves the problem that the traditional intersection collision avoidance warning method based on trajectory prediction fails to consider the severity of the conflict and the comprehensive risk, and the warning strategy has poor reliability.
[0006] (II) Technical solution
[0007] To achieve the above object, the present invention provides the following technical solution: a collision avoidance warning method at an intersection based on trajectory prediction, comprising the following steps:
[0008] Step 1: The data acquisition module obtains the image data of the driving trajectories of all traffic participants at the intersection in real time through a network connection monitoring device, and forms a trajectory data set according to the screening criteria;
[0009] Step 2: The data acquisition module is equipped with LSTM and GRU. The data acquisition module substitutes the trajectory data set into LSTM for encoding to obtain the latent space feature vector, and then uses LSTM and GRU to decode the latent space feature vector, analyzes and obtains the predicted trajectory of all traffic participants in the next 8 seconds, and forms a predicted data set;
[0010] Step 3: The conflict detection module calculates and generates the trajectory overlap point Chd based on the predicted data set, analyzes the data of traffic participants whose trajectories intersect, and then determines whether there is a conflict point in the trajectory overlap point Chd after a time limit. If there is no conflict point in the trajectory overlap point Chd, no further analysis is required;
[0011] Step 4: If it is determined in step 3 that there is a conflict point in the trajectory overlap point Chd, the conflict detection module calculates the collision arrival time TTC based on the trajectory overlap point Chd and the predicted data set;
[0012] Step 5: The anti-collision warning module calculates and generates the warning threshold T based on the trajectory data set, the prediction data set and the trajectory coincidence point Chd, then compares the collision arrival time TTC with the warning threshold Yjyz, and outputs the corresponding warning signal.
[0013] Preferably, in step 1, the screening criteria require retaining the 7 frames of image data of all traffic participants at the intersection at the most recent time point, and the trajectory data set includes the 7 frames of image data of all traffic participants at the intersection at the most recent time point. The expression of the trajectory data set is: to The 7 frames of image data corresponding to the latest time point of each traffic participant in the intersection in sequence include the shape features, position features and movement features of the traffic participant. The shape features include length, width and height. The position features include two-dimensional coordinates and a direction scalar. The movement features include speed, acceleration and direction acceleration. 7t represents the time point corresponding to the 7 frames of image data. The subscripts 1 to n represent that there are n traffic participants in the intersection.
[0014] Preferably, in step 2, the prediction data set includes the predicted trajectories of all traffic participants at the intersection in the next 8 seconds, and the expression of the prediction data set is: to The predicted trajectory of each traffic participant in the intersection in the next 8s corresponds to the predicted trajectory of each traffic participant in the next 8s. The predicted trajectory of each traffic participant in the next 8s includes the shape characteristics, position characteristics and running characteristics of the traffic participant. The shape characteristics include length, width and height, the position characteristics include two-dimensional coordinates and orientation scalars, and the motion characteristics include speed, acceleration and orientation acceleration. 8s represents the time point corresponding to the predicted trajectory in the next 8s. The subscripts 1 to n represent that there are n traffic participants in the intersection. The trajectory dataset is and prediction data set They all correspond to the same traffic participant.
[0015] Preferably, in step 3, the calculation formula of the trajectory coincidence point Chd is as follows:
[0016]
[0017] In the formula, Chd represents the trajectory coincidence point, and To predict any two traffic participants in the data set, represents the predicted trajectory of the i-th future 8s in the prediction data set, represents the predicted trajectory of the kth future 8s in the prediction data set, It means that the distance difference between the predicted trajectories of any two traffic participants in the prediction data set in the next 8 seconds is calculated at the same time. If the difference between the two distances is less than or equal to 0.5 meters, it is the data of the traffic participant whose trajectories intersect.
[0018] Preferably, in the step three, a time point judgment is performed on the data of traffic participants whose trajectories intersect at the trajectory overlap point Chd, based on the predicted trajectory start time point. If the time point difference between any two traffic participants is greater than 8S, the conflict detection module determines that there is no conflict point at the trajectory overlap point Chd, and no further analysis is required.
[0019] Preferably, in step three, if the time point difference between any two traffic participants is less than or equal to 8S, the conflict detection module determines that there is a conflict point at the trajectory overlap point Chd.
[0020] Preferably, in step 4, the collision arrival time TTC calculation formula is as follows:
[0021]
[0022] In the formula, TTC represents the collision arrival time, Dre l represents the relative distance between any two traffic participants, D0 represents the constant in the collision arrival time calculation formula, that is, the safe stopping distance, V rel represents the relative speed of any two traffic participants, a rel Represents the relative acceleration of any two traffic participants.
[0023] Preferably, in step 5, the calculation process of the warning threshold T is as follows:
[0024] T=ty+tf+tz+tj+ts
[0025] In the formula, T represents the warning threshold, ty, tf, tz and tj are fixed parameters in the calculation formula of the warning threshold, ty represents the network communication delay time, tf represents the reaction time of the traffic participant to the warning signal, tz represents the brake coordination time, tj represents the time when the braking acceleration reaches the maximum, and ts represents the time when the traffic participant stops when the maximum braking acceleration is reached;
[0026] The time ts for a traffic participant to stop when reaching the maximum braking acceleration is calculated as follows:
[0027] According to the trajectory data set and the prediction data set, the initial velocity of the traffic participant corresponding to the conflict point Chd of the trajectory coincidence point is marked as v, the acceleration is marked as a, and the maximum braking acceleration is marked as a0;
[0028] Calculate the displacement distance S of traffic participants in the time period tj tj , which is expressed as follows:
[0029]
[0030] When the traffic participant reaches the maximum braking speed, it will continue to decelerate according to the maximum braking speed, so it is a uniform deceleration motion at this time, until the speed of the traffic participant is 0, and the displacement of the car in time ts is set to S ts , S is the distance from the car to the collision point, and its expression is as follows:
[0031] S ts =S-(ty+tf+tz)vS tj
[0032] After tj, the speed of the traffic participant becomes v0, which is expressed as follows:
[0033]
[0034] Integrating the above formula according to the distance formula,
[0035]
[0036] Preferably, in step five, if the traffic participants corresponding to the trajectory coincidence point Chd and the conflict point are all within the intersection, and the collision arrival time TTC is less than or equal to the warning threshold T, it means that the traffic participants are about to collide, and the anti-collision warning module outputs an emergency collision warning signal.
[0037] Preferably, in step five, if the traffic participants corresponding to the trajectory overlap point Chd and the conflict point are all within the intersection, and the collision arrival time TTC is greater than the warning threshold T, it means that there is a conflict object in the predicted trajectory of the traffic participant's location in the next 8 seconds, and the anti-collision warning module outputs a prompt warning signal.
[0038] Compared with the prior art, the present invention provides a crossroads anti-collision warning method based on trajectory prediction, which has the following beneficial effects:
[0039] 1. The present invention connects the monitoring device through the data acquisition module to obtain the image data of the driving trajectory of all traffic participants in the intersection in real time, and forms a trajectory data set according to the screening criteria. The data acquisition module is provided with LSTM and GRU. The data acquisition module substitutes the trajectory data set into LSTM for encoding to obtain the latent space feature vector, and then uses LSTM and GRU to decode the latent space feature vector, analyzes and obtains the predicted trajectory of all traffic participants in the next 8s, and forms a predicted data set. The conflict detection module calculates and generates the trajectory overlap point Chd according to the predicted data set. If the difference between the two distances is less than or equal to 0.5 meters, it is the data of the traffic participants whose trajectories intersect, and the prediction trajectory conflict has high accuracy.
[0040] 2. The present invention judges the time point of the traffic participant data whose trajectories intersect in the trajectory overlap point Chd, and takes the predicted trajectory starting time point as the standard. If the time point difference between any two traffic participants is greater than 8S, the conflict detection module judges that there is no conflict point at the trajectory overlap point Chd, and no further analysis is required. If the time point difference between any two traffic participants is less than or equal to 8S, the conflict detection module judges that there is a conflict point at the trajectory overlap point Chd. The conflict detection module calculates the collision arrival time TTC based on the trajectory overlap point Chd and the predicted data set. The anti-collision warning module calculates and generates the warning threshold T based on the trajectory data set, the predicted data set and the trajectory overlap point Chd, and then compares the collision arrival time TTC with the warning threshold Yjyz, and outputs the corresponding warning signal. The two-level warning method can remind the driver of the current traffic conditions and possible collisions, and the two-level warning driving safety is high. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 It is a step diagram of the method of the present invention. DETAILED DESCRIPTION
[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0043] Since the traditional intersection collision avoidance warning method based on trajectory prediction fails to consider the severity of the conflict and the comprehensive risk, the existing driving safety data is derived from the line of sight of a single vehicle body. The data source is single, and a single vehicle body cannot obtain more comprehensive driving safety information other than the vehicle itself. In the actual warning process, the warning method does not take into account the characteristics of the driver, and the reliability of the warning strategy is poor. Therefore, a crossroad collision avoidance warning method based on trajectory prediction is provided. Please refer to Figure 1 , the method comprises the following steps:
[0044] Step 1: The data acquisition module obtains the image data of the driving trajectory of all traffic participants in the intersection in real time through the network connection monitoring device, and forms a trajectory data set according to the screening criteria. The screening criteria require retaining the 7 frames of image data of all traffic participants in the intersection at the most recent time point. The trajectory data set includes the 7 frames of image data of all traffic participants at the most recent time point at the intersection. The expression of the trajectory data set is: to The 7 frames of image data corresponding to the most recent time point of each traffic participant in the intersection are sequentially obtained. The 7 frames of image data of each traffic participant at the most recent time point include the shape features, position features and running features of the traffic participant. The shape features include length, width and height, the position features include two-dimensional coordinates and orientation scalars, and the motion features include speed, acceleration and orientation acceleration. The driving trajectory features of all traffic participants are comprehensively analyzed to facilitate the subsequent use of historical data features for more accurate trajectory prediction. 7t represents the time point corresponding to the 7 frames of image data, and the subscripts 1 to n represent that there are n traffic participants in the intersection.
[0045] Step 2: The data acquisition module is equipped with LSTM and GRU. The data acquisition module substitutes the trajectory data set into LSTM for encoding to obtain the latent space feature vector, and then uses LSTM and GRU to decode the latent space feature vector, analyzes and obtains the predicted trajectories of all traffic participants in the next 8 seconds, and forms a predicted data set. The predicted data set includes the predicted trajectories of all traffic participants in the intersection in the next 8 seconds. The expression of the predicted data set is to The predicted trajectory of each traffic participant in the intersection in the next 8s corresponds to the predicted trajectory of each traffic participant in the next 8s. The predicted trajectory of each traffic participant in the next 8s includes the shape characteristics, position characteristics and running characteristics of the traffic participant. The shape characteristics include length, width and height, the position characteristics include two-dimensional coordinates and orientation scalars, and the motion characteristics include speed, acceleration and orientation acceleration. 8s represents the time point corresponding to the predicted trajectory in the next 8s. The subscripts 1 to n represent that there are n traffic participants in the intersection. The trajectory dataset is and prediction data set They all correspond to the same traffic participant;
[0046] Step 3: The conflict detection module calculates and generates the trajectory overlap point Chd based on the predicted data set, analyzes the data of traffic participants whose trajectories intersect, and then determines whether there is a conflict point in the trajectory overlap point Chd after a time limit. If there is no conflict point in the trajectory overlap point Chd, no further analysis is required;
[0047] The calculation formula of the trajectory coincidence point Chd is as follows:
[0048]
[0049] In the formula, Chd represents the trajectory coincidence point, and To predict any two traffic participants in the data set, represents the predicted trajectory of the i-th future 8s in the prediction data set, represents the predicted trajectory of the kth future 8s in the prediction data set, It means that the distance difference between the predicted trajectories of any two traffic participants in the prediction data set in the next 8 seconds is calculated at the same time. If the difference between the two distances is less than or equal to 0.5 meters, it is the data of the traffic participants whose trajectories intersect, and the prediction trajectory conflict accuracy is high;
[0050] The time point of the traffic participants whose trajectories intersect at the trajectory overlap point Chd is judged based on the predicted trajectory start time point. If the time point difference between any two traffic participants is greater than 8S, the conflict detection module determines that there is no conflict point at the trajectory overlap point Chd and no further analysis is required. If the time point difference between any two traffic participants is less than or equal to 8S, the conflict detection module determines that there is a conflict point at the trajectory overlap point Chd.
[0051] Step 4: If the collision point exists in the trajectory overlap point Chd determined in step 3, the collision detection module calculates the collision arrival time TTC according to the trajectory overlap point Chd and the predicted data set. The calculation formula is as follows:
[0052]
[0053] In the formula, TTC represents the collision arrival time, D rel represents the relative distance between any two traffic participants, D0 represents the constant in the collision arrival time calculation formula, that is, the safe stopping distance, V rel represents the relative speed of any two traffic participants, a rel represents the relative acceleration of any two traffic participants;
[0054] Step 5: The anti-collision warning module calculates and generates the warning threshold T according to the trajectory data set, the prediction data set and the trajectory coincidence point Chd, then compares the collision arrival time TTC with the warning threshold Yjyz, and outputs the corresponding warning signal;
[0055] The calculation process of the warning threshold T is as follows:
[0056] T=ty+tf+tz+tj+ts
[0057] In the formula, T represents the warning threshold, ty, tf, tz and tj are fixed parameters in the calculation formula of the warning threshold, ty represents the network communication delay time, that is, the time for the vehicle to receive various information after calculation from the background to the vehicle, which can be set to 0.1s, tf represents the reaction time of the traffic participant to the warning signal, which can be set to 0.3s-1s, tz represents the brake coordination time, that is, the time from the driver preparing to brake to the start of braking, which can be set to 0.3s, tj represents the time when the braking acceleration reaches the maximum, which can be set to 0.45s, and ts represents the time when the traffic participant stops when the maximum braking acceleration is reached;
[0058] The time ts for a traffic participant to stop when reaching the maximum braking acceleration is calculated as follows:
[0059] According to the trajectory data set and the prediction data set, the initial velocity of the traffic participant corresponding to the conflict point Chd of the trajectory coincidence point is marked as v, the acceleration is marked as a, and the maximum braking acceleration is marked as a0;
[0060] Calculate the displacement distance S of traffic participants in the time period tj tj , which is expressed as follows:
[0061]
[0062] When the traffic participant reaches the maximum braking speed, it will continue to decelerate according to the maximum braking speed, so it is a uniform deceleration motion at this time, until the speed of the traffic participant is 0, and the displacement of the car in time ts is set to S ts , S is the distance from the car to the collision point, and its expression is as follows:
[0063] S ts =S-(ty+tf+tz)vS tj
[0064] After tj, the speed of the traffic participant becomes v0, which is expressed as follows:
[0065]
[0066] Integrating the above formula according to the distance formula,
[0067]
[0068] If the traffic participants corresponding to the conflict points of the trajectory overlap point Chd are all within the intersection, and the collision arrival time TTC is less than or equal to the warning threshold T, it means that the traffic participants are about to collide. The anti-collision warning module outputs an emergency collision warning signal. If the traffic participants corresponding to the conflict points of the trajectory overlap point Chd are all within the intersection, and the collision arrival time TTC is greater than the warning threshold T, it means that there is a conflict object in the predicted trajectory of the traffic participant in the next 8 seconds. The anti-collision warning module outputs a prompt warning signal. The traffic conditions at the intersection are complicated. For example, there are often blind spots in the field of vision of traffic participants. The two-level warning method can remind the driver of the current traffic conditions and possible collisions. The two-level warning has high driving safety.
[0069] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A collision avoidance warning method for intersections based on trajectory prediction, characterized in that: The following steps are involved: Step 1: The data acquisition module obtains the image data of the driving trajectories of all traffic participants at the intersection in real time through a network connection monitoring device, and forms a trajectory data set according to the screening criteria; Step 2: The data acquisition module is equipped with LSTM and GRU. The data acquisition module substitutes the trajectory data set into LSTM for encoding to obtain the latent space feature vector, and then uses LSTM and GRU to decode the latent space feature vector, analyzes and obtains the predicted trajectory of all traffic participants in the next 8 seconds, and forms a predicted data set; Step 3: The conflict detection module calculates and generates the trajectory overlap point Chd based on the predicted data set, analyzes the data of traffic participants whose trajectories intersect, and then determines whether there is a conflict point in the trajectory overlap point Chd after a time limit. If there is no conflict point in the trajectory overlap point Chd, no further analysis is required; Step 4: If it is determined in step 3 that there is a conflict point in the trajectory overlap point Chd, the conflict detection module calculates the collision arrival time TTC based on the trajectory overlap point Chd and the predicted data set; Step 5: The anti-collision warning module calculates and generates the warning threshold T based on the trajectory data set, the prediction data set and the trajectory coincidence point Chd, then compares the collision arrival time TTC with the warning threshold Yjyz, and outputs the corresponding warning signal.
2. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 1, characterized in that: In step 1, the screening criteria require retaining the 7 frames of image data of all traffic participants at the intersection at the most recent time point. The trajectory data set includes the 7 frames of image data of all traffic participants at the intersection at the most recent time point. The expression of the trajectory data set is: to The 7 frames of image data corresponding to the latest time point of each traffic participant in the intersection in sequence include the shape features, position features and movement features of the traffic participant. The shape features include length, width and height. The position features include two-dimensional coordinates and a direction scalar. The movement features include speed, acceleration and direction acceleration. 7t represents the time point corresponding to the 7 frames of image data. The subscripts 1 to n represent that there are n traffic participants in the intersection.
3. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 2, characterized in that: In step 2, the prediction data set includes the predicted trajectories of all traffic participants in the intersection in the next 8 seconds. The expression of the prediction data set is: to The predicted trajectory of each traffic participant in the intersection in the next 8s corresponds to the predicted trajectory of each traffic participant in the next 8s. The predicted trajectory of each traffic participant in the next 8s includes the shape characteristics, position characteristics and running characteristics of the traffic participant. The shape characteristics include length, width and height, the position characteristics include two-dimensional coordinates and orientation scalars, and the motion characteristics include speed, acceleration and orientation acceleration. 8s represents the time point corresponding to the predicted trajectory in the next 8s. The subscripts 1 to n represent that there are n traffic participants in the intersection. The trajectory dataset is and prediction data set They all correspond to the same traffic participant.
4. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 3, characterized in that: In step 3, the calculation formula of the trajectory coincidence point Chd is as follows: In the formula, Chd represents the trajectory coincidence point, and To predict any two traffic participants in the data set, represents the predicted trajectory of the i-th future 8s in the prediction data set, represents the predicted trajectory of the kth future 8s in the prediction data set, It means that the distance difference between the predicted trajectories of any two traffic participants in the prediction data set in the next 8 seconds is calculated at the same time. If the difference between the two distances is less than or equal to 0.5 meters, it is the data of the traffic participant whose trajectories intersect.
5. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 4, characterized in that: In the step three, a time point judgment is performed on the data of traffic participants whose trajectories intersect at the trajectory overlap point Chd, based on the predicted trajectory start time point. If the time point difference between any two traffic participants is greater than 8S, the conflict detection module determines that there is no conflict point at the trajectory overlap point Chd, and no further analysis is required.
6. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 5, characterized in that: In the step 3, if the time difference between any two traffic participants is less than or equal to 8S, the conflict detection module determines that there is a conflict point at the trajectory overlap point Chd.
7. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 6, characterized in that: In step 4, the collision arrival time TTC is calculated as follows: In the formula, TTC represents the collision arrival time, D rel represents the relative distance between any two traffic participants, D0 represents the constant in the collision arrival time calculation formula, that is, the safe stopping distance, V rel represents the relative speed of any two traffic participants, a rel Represents the relative acceleration of any two traffic participants.
8. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 7, characterized in that: In step 5, the calculation formula of the warning threshold T is as follows: T=ty+tf+tz+tj+ts In the formula, T represents the warning threshold, ty, tf, tz and tj are fixed parameters in the calculation formula of the warning threshold, ty represents the network communication delay time, tf represents the reaction time of the traffic participant to the warning signal, tz represents the brake coordination time, tj represents the time when the braking acceleration reaches the maximum, and ts represents the time when the traffic participant stops when the maximum braking acceleration is reached; The time ts for a traffic participant to stop when reaching the maximum braking acceleration is calculated as follows: According to the trajectory data set and the prediction data set, the initial velocity of the traffic participant corresponding to the conflict point Chd of the trajectory coincidence point is marked as v, the acceleration is marked as a, and the maximum braking acceleration is marked as a0; Calculate the displacement distance S of traffic participants in the time period tj tj , which is expressed as follows: When the traffic participant reaches the maximum braking speed, it will continue to decelerate according to the maximum braking speed, so it is a uniform deceleration motion at this time, until the speed of the traffic participant is 0, and the displacement of the car in time ts is set to S ts , S is the distance from the car to the collision point, and its expression is as follows: With ts =S-(ty+tf+tz)vS tj After tj, the speed of the traffic participant becomes v0, which is expressed as follows: Integrating the above formula according to the distance formula, 9. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 8, characterized in that: In step five, if the traffic participants corresponding to the trajectory coincidence point Chd and the conflict point are all within the intersection, and the collision arrival time TTC is less than or equal to the warning threshold T, it means that the traffic participants are about to collide, and the anti-collision warning module outputs an emergency collision warning signal.
10. The method for anti-collision warning at an intersection based on trajectory prediction according to claim 9, characterized in that: In step five, if the traffic participants corresponding to the trajectory overlap point Chd and the conflict point are all within the intersection, and the collision arrival time TTC is greater than the warning threshold T, it means that there is a conflict object in the predicted trajectory of the traffic participant in the next 8 seconds, and the anti-collision warning module outputs a prompt warning signal.
