A vehicle guidance method in long tunnels based on radar-visual fusion data

Abstract

This invention discloses a vehicle guidance method for long tunnels based on radar-visual fusion data. This method considers the impact of slow-moving vehicles on actual road traffic, analyzes vehicle trajectory information to identify multiple slow-moving vehicles, and predicts two-lane travel times based on historical slow-moving vehicle trajectory data. By comparing the travel times of the two lanes within the tunnel and displaying them on variable information boards at the tunnel entrance, it guides vehicles to change lanes and provide speed guidance. This method is simple and easy to implement, exhibits excellent real-time performance and adaptability, and can improve lane utilization and enhance road traffic efficiency.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent transportation, and in particular relates to a vehicle guidance method in a long tunnel based on radar-visual fusion data. Background Art

[0002] With the development of intelligent transportation systems, radar-visual fusion technology has emerged, revolutionizing traffic perception. Combining the strengths of radar and video sensors, it provides a variety of accurate, efficient, and real-time traffic perception data, including real-time vehicle trajectory data. Considering the complex operating conditions of long and high-speed tunnels, high vehicle speeds, and limited driver vision, tunnel guidance methods based on radar-visual fusion data are needed to rationally distribute traffic flow, reduce congestion and traffic accidents, and improve traffic efficiency.

[0003] Existing research focuses on variable message board speed guidance and collaborative speed and lane change guidance in intelligent connected environments. For example, Papamichail et al. designed a controller based on a dual-loop feedback cascade control structure to adjust the speed limit. This controller uses the difference between the real-time traffic density occupancy and the desired density occupancy to control upstream speed. However, the macro-estimated variable speed limit fails to consider factors such as actual traffic conditions, vehicle trajectories, and real-time information, making it impractical for vehicle guidance in actual highway tunnel scenarios. Jiao Pengpeng et al. proposed a method for lane change guidance for bottleneck vehicles on highways in a cooperative vehicle-infrastructure environment. They established utility functions for both free and forced lane change behaviors, guiding upstream vehicles through the cooperative vehicle-infrastructure system to improve lane utilization. However, the level of intelligent connectedness and CAV penetration in existing highway tunnels is low, and the experimental scenarios used in their method may not match actual conditions, resulting in insufficient applicability and a failure to meet practical application requirements. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a vehicle guidance method for long tunnels based on radar-visual fusion data. The present invention aims to solve the problems of traditional roadside perception technology, such as low real-time performance, difficult fusion, and large prediction errors.

[0005] To achieve the above object, the present invention provides a method for guiding vehicles in long tunnels based on radar-visual fusion data, comprising the following steps:

[0006] S1. Detect vehicles passing through the tunnel entrance using roadside equipment installed at the tunnel entrance, record the time t when the target vehicle passes through the tunnel entrance, and obtain radar-visual fusion information for each lane in the tunnel at time t;

[0007] The radar-visual fusion information is the target data of all traffic participants in each frame in the tunnel;

[0008] S2. Based on the radar fusion information collected in step S1, the vehicle trajectory data in the tunnel is smoothed;

[0009] S3. Traverse the vehicle list information List1 on the lanes in the tunnel and obtain the speed V of each vehicle i and position X i , by comparing the speeds, find the slowest vehicle in the target lane at time t slow1 , and obtain the speed V of the slowest vehicle slow1 and position X slow1 ;

[0010] S4. Vehicle-based slow1 Historical trajectory data, select the linear regression algorithm to predict the estimated Vehicle slow1 Time to arrive at the tunnel exit, that is, Vehicle slow1 Travel time T slow1 ;

[0011] S5. When predicting Vehicle slow1 When exiting the tunnel, turn on Vehicle slow1 All vehicles between the last vehicle in the target lane in the tunnel are taken as the new vehicle list information List2. Repeat step S3 to obtain the second slowest vehicle in the target lane Vehicle slow2 Speed ​​V slow2 and position X slow2 Repeat step S4 to obtain T through linear regression slow1 InsideVehicle slow2 The distance traveled, then repeat step S4 to predict the vehicle slow2 Time to arrive at the tunnel exit, that is, Vehicle slow2 Travel time T slow2 ;

[0012] S6. Based on steps S3, S4, and S5, predict the travel time of the target lane in the tunnel at time t;

[0013] S7. Vehicle Guidance

[0014] The travel time of each lane in the tunnel at time t is obtained in step S6, and the travel time of each lane at time t is displayed on the variable information board at the tunnel entrance area, thereby guiding vehicles to change lanes;

[0015] When subsequent vehicles enter the tunnel entrance, the above steps are repeated to predict the travel time of each lane in the tunnel, thereby guiding the vehicle to change lanes.

[0016] Furthermore, in step S1, the data frame rate of the radar-visual fusion information is 10 frames per second.

[0017] Furthermore, in step S1, the traffic participant target data includes: ① vehicle id: Vehicle i ;②Lane number lanenum:lane i ; ③ Speed ​​(km / h): V i ④Position x(m):X i .

[0018] Furthermore, step S2 includes the following sub-steps:

[0019] S2.1 sets the attenuation coefficient α and uses the first-order inertial filter to filter the vehicle i The trajectory data is smoothed every 0.1s;

[0020] The filtering calculation formula is as follows:

[0021] smooth_speed[i]=α*V[i]+(1-α)*smooth_speed[i-1]

[0022] smoothed_speed[0] = V[0]

[0023] Where α represents the filter coefficient; smooth_speed[i] represents the output value of the current filtering speed; smooth_speed[i-1] represents the output value of the previous filtering speed; V[i] represents the current sampling speed value; V[0] represents the first frame speed;

[0024] S2.2 removes the highest speed V among the 10 filtered speed output values ​​within 1s fast and the minimum speed V slow , using the median average method, the average value of the remaining 8 filtered speed output values ​​is obtained as Vehicle i Velocity information V at 1s intervals i .

[0025] Furthermore, step S4 includes the following sub-steps:

[0026] S4.1 Select Vehicle slow1 n sets of data before time t, including time t (t n ,t n-1 ,....t1) and the position x(x n ,x n-1 ,....x1), build the relationship between time and position:

[0027] x=at+b

[0028] Where x is the position, representing the dependent variable to be predicted; t is the time, representing the independent variable to be predicted; a is the slope of the regression line; b is a constant;

[0029] S4.2 Solve for the values ​​of a and b:

[0030] S4.3 is predicted to be located at Xt slow1 The slowest vehicle slow1 Travel time to the tunnel exit T slow1 , the formula is as follows:

[0031]

[0032] Where Length represents the tunnel length.

[0033] Furthermore, step S6 includes the following sub-steps:

[0034] S6.1 Repeat steps S3, S4, and S5, iteratively find the slow-moving vehicles in the target lane vehicle list and make predictions to obtain the travel time T slow1 、T slow2 .....T slown , until the last vehicle in the target lane is found last until;

[0035] S6.2 Get the last vehicle in the target lane last Position X last and Vehicle last From the entrance to X last Time T last ;

[0036] S6.3 Predict the travel time of the target lane in the tunnel at time t using the following formula:

[0037] T pre =T slow1 +T slow2 +…+T slown +T last

[0038] Where, T pre represents the travel time of the target lane in the tunnel at time t.

[0039] The beneficial effects of the present invention are:

[0040] The present invention discloses a vehicle guidance method for long tunnels based on radar-visual fusion data. The method takes into account the impact of slow-moving vehicles on vehicle operation in actual traffic roads, analyzes vehicle trajectory information to identify multiple slow-moving vehicles, and predicts the travel time of two lanes based on the historical trajectory data of slow-moving vehicles. By comparing the travel time of the two lanes in the tunnel and displaying the travel time of the two lanes on the variable information board at the tunnel entrance area, the method guides vehicles to change lanes and provide speed guidance. The method is simple and easy to implement, has good real-time and adaptability, can improve lane utilization, and enhance road traffic efficiency.

[0041] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a flow chart of a vehicle guidance method for long tunnels based on radar-visual fusion data;

[0043] Figure 2 Schematic diagram of vehicles driving in a two-lane tunnel. DETAILED DESCRIPTION

[0044] To make the technical solutions, advantages, and purposes of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described 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 this application.

[0045] like Figure 1 As shown, the present invention provides a vehicle guidance method for a long tunnel based on radar-visual fusion data, comprising:

[0046] S1. Detect vehicles passing through the tunnel entrance using roadside equipment installed at the tunnel entrance, record the time t when the target vehicle passes through the tunnel entrance, and obtain radar-visual fusion information for each lane in the tunnel at time t;

[0047] The fusion of radar and vision can obtain the precise location and vectorized data of each vehicle. The fusion of radar and vision information is the target data of all traffic participants in each frame in the tunnel, with a data frame rate of 10 frames per second. Among them, the target data of traffic participants includes: ① Vehicle id: Vehicle i ;②Lane number lanenum:lane i ; ③ Speed ​​(km / h): Vi ④Position x(m):X i .

[0048] S2. Based on the radar fusion information collected in step S1, the vehicle trajectory data in the tunnel is smoothed;

[0049] The visual fusion data frame rate is 10 frames per second. The vehicle speed data output every 0.1 seconds may have large fluctuations. Therefore, the vehicle trajectory data needs to be smoothed. Here, the vehicle speed data of 10 frames per second is taken as an example.

[0050] S2.1 uses a first-order inertial filter to filter these 10 data. First, the smoothing result is initialized to the first sample value, that is, smoothed_speed[0]=V[0], where V[0] represents the first frame speed.

[0051] Then apply the following formula to filter its velocity:

[0052] smooth_speed[i]=α*V[i]+(1-α)*smooth_speed[i-1]

[0053] Where α is the filter coefficient (here α = 0.2); smooth_speed[i] is the current filtered speed output; smooth_speed[i-1] is the previous filtered speed output; and V[i] is the current sampled speed value. The current sampled value is weighted with the previous filtered output value to obtain the effective speed filter value, so that the output has a feedback effect on the input.

[0054] S2.2 uses the median average method to remove the highest speed V among the 10 data fast and the minimum speed V slow , calculate the average value of the remaining 8 data as Vehicle i Velocity information V at 1s intervals i :

[0055]

[0056] S3. Traverse the vehicle list information List1 on the lanes in the tunnel and obtain the speed V of each vehicle i and position X i , by comparing the speeds, find the slowest vehicle in the target lane at time t slow1 , and obtain the speed V of the slowest vehicle slow1 and position X slow1 ;

[0057] like Figure 2As shown, the tunnel in this embodiment is a two-lane tunnel. The vehicle list List1 of the two lanes lane0 and lane1 is obtained by traversing. Here, lane lane0 in the tunnel is taken as an example, where the number of vehicles is n. The vehicle list obtained is:

[0058] List1=[Vehicle n ,Vehicle n-1 ,....,Vehicle3,Vehicle2,Vehicle1]

[0059] The order of the vehicle list from left to right is the order of vehicles traveling in the tunnel, that is, Vehicle n Vehicle 1 represents the vehicle closest to the tunnel entrance, and Vehicle 2 represents the vehicle closest to the tunnel exit.

[0060] Traverse the vehicle list List1 and get the speed V of each vehicle i , and compare the vehicle speeds to get the slowest vehicle slow1 :

[0061] V slow1 ≤V[i],i=1,2...n

[0062] Among them, V slow1 Indicates the slowest vehicle Vehicle slow1 At the same time, based on the Vehicle slow1 Vehicle id, you can get Vehicle slow1 Position X at time t slow1 , taking the tunnel entrance as the coordinate origin, here we only analyze the longitudinal position of the vehicle in the tunnel.

[0063] S4. Vehicle-based slow1 Historical trajectory data, select the linear regression algorithm to predict the estimated Vehicle slow1 Time to arrive at the tunnel exit, that is, Vehicle slow1 Travel time T slow1 ;

[0064] First, select Vehicle slow1 If there are n groups of data before time t, then time t is selected as the independent variable (t(t n ,t n-1 ,....t1), select position x(x n ,x n-1 ,...x1) as the dependent variable, assuming the relationship between them is: x = at + b. Where x is the position of the dependent variable being predicted, t is the time of the independent variable being predicted, a is the slope of the regression line, and b is a constant.

[0065] Secondly, in order to measure the degree of difference between the model prediction value and the true value, the error square sum SSE is used as the loss function:

[0066]

[0067] Among them, L(a,b) represents the sum of squared errors, f(t i ) represents the predicted position at time i.

[0068] Then, to obtain the best a and b, the least squares method is used to find the best function matching the data by minimizing the sum of squared errors. Therefore, it is necessary to minimize the loss function L(a,b), which can be obtained by taking the partial derivative of L(a,b) and making the first-order derivative equal to 0:

[0069]

[0070] At this time, L(x,f(t)) reaches the minimum value, and the solution is:

[0071]

[0072] Substitute n sets of data into the formula to obtain the values ​​of a and b, and then obtain the linear equation x=at+b.

[0073] Finally, the prediction is slow1 The slowest vehicle slow1 When arriving at the tunnel exit, the travel time is:

[0074]

[0075] Among them, T slow1 Represents Vehicle slow1 The estimated travel time to the tunnel exit. Length indicates the length of the tunnel.

[0076] S5. When predicting Vehicle slow1 When exiting the tunnel, the vehicle list information in the tunnel needs to be reselected:

[0077] List2=[Vehicle n ,Vehicle n-1 ,....,Vehicle slow1+2 ,Vehicle slow1+1 ]

[0078] That is, you need to use Vehicle slow1 All vehicles between the last vehicle in the target lane in the tunnel are taken as the new vehicle list information List2;

[0079] Repeat step S3 to find the slowest vehicle in the target lane in the vehicle list information List2. slow2 ; and get the slowest vehicle Vehicle slow2 Speed ​​V slow2 and position X slow2 .

[0080] At this time, you need to consider the Vehicle slow1 Travel time T slow1 InsideVehicle slow2 The distance traveled is repeated in step S4, and T is obtained by linear regression. slow1 InsideVehicle slow2 Distance traveled, Vehicle slow2 The distance to the tunnel exit is: Length-X slow2 -X Tslow1 Then repeat step S4 to predict Vehicle slow2 In X slow2 +X Tslow1 The time it takes to exit the tunnel at the location, that is, the vehicle slow2 Travel time T slow2 ;

[0081] S6. Based on steps S3, S4, and S5, predict the travel time of the target lane in the tunnel at time t;

[0082] S6.1 Repeat steps S3, S4, and S5 above, find the slow-moving vehicles in the target lane in the vehicle list and make predictions, then iteratively reduce the range of the vehicle list until the last vehicle in the target lane in the tunnel is found. n until:

[0083] List2=[Vehicle n ,Vehicle n-1 ,....,Vehicle slow2+2 ,Vehicle slow2+1 ]

[0084] List3=[Vehicle n ,Vehicle n-1 ,....,Vehicle slow3+2 ,Vehicle slow3+1 ] ......

[0086] List i =[Vehicle n ]

[0087] By finding the slow-moving vehicles in each vehicle list, and then repeating step S4 to predict the travel time T slow1 、T slow2 …T slown .

[0088] The last vehicle in the target lane in the tunnel has entered the tunnel at time t. Get Vehicle n From the entrance to X slown Time T last .

[0089] Finally, the travel time of lane 0 in the tunnel at time t is obtained as:

[0090] T lane0-traveltime =T slow1 +T slow2 +...+T slown +T last

[0091] Similarly, the travel time T of lane 1 in the tunnel at time t can be obtained lane1-traveltime .

[0092] S7. Vehicle Guidance

[0093] Based on the above steps, the travel time T of the two lanes in the tunnel at time t is obtained. lane0-traveltime and T lane1-traveltime , displaying the two-trip time on the variable information board at the tunnel entrance area, guiding vehicles to change lanes and speed, improving lane utilization and road traffic efficiency;

[0094] When a new vehicle enters the tunnel entrance at time t+i, the above steps are repeated to predict the tunnel dual-lane travel time, which is displayed on the variable information board to guide the driver to change lanes.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of protection of the present invention.

Claims

1. A vehicle guidance method for long tunnels based on radar-visual fusion data, characterized in that: The following steps are involved: S1. Detect vehicles passing through the tunnel entrance using roadside equipment installed at the tunnel entrance, record the time t when the target vehicle passes through the tunnel entrance, and obtain radar-visual fusion information for each lane in the tunnel at time t; The radar-visual fusion information is the target data of all traffic participants in each frame in the tunnel; S2. Based on the radar fusion information collected in step S1, the vehicle trajectory data in the tunnel is smoothed; S3. Traverse the vehicle list information List1 on the lanes in the tunnel and obtain the speed V of each vehicle i and position X i , by comparing the speeds, find the slowest vehicle in the target lane at time t slow1 , and obtain the speed V of the slowest vehicle slow1 and position X slow1 ; S4. Vehicle-based slow1 Historical trajectory data, select the linear regression algorithm to predict the estimated Vehicle slow1 Time to arrive at the tunnel exit, that is, Vehicle slow1 Travel time T slow1 ; S5. When predicting Vehicle slow1 When exiting the tunnel, turn on Vehicle slow1 All vehicles between the last vehicle in the target lane in the tunnel are taken as the new vehicle list information List2. Repeat step S3 to obtain the second slowest vehicle in the target lane Vehicle slow2 Speed ​​V slow2 and position X slow2 Repeat step S4 to obtain T through linear regression slow1 InsideVehicle slow2 The distance traveled, then repeat step S4 to predict the vehicle slow2 Time to arrive at the tunnel exit, that is, Vehicle slow2 Travel time T slow2 ; S6. Based on steps S3, S4, and S5, predict the travel time of the target lane in the tunnel at time t; S6.1 Repeat steps S3, S4, and S5, iteratively find the slow-moving vehicles in the target lane vehicle list and make predictions to obtain the travel time T slow1 、T slow2 …T slown , until the last vehicle in the target lane is found last until; S6.2 Get the last vehicle in the target lane last Position X last and Vehicle last From the entrance to X last Time T last ; S6.3 Predict the travel time of the target lane in the tunnel at time t using the following formula: T pre =T slow1 +T slow2 +…+T slown +T last Where, T pre represents the travel time of the target lane in the tunnel at time t; S7. Vehicle Guidance The travel time of each lane in the tunnel at time t is obtained in step S6, and the travel time of each lane at time t is displayed on the variable information board at the tunnel entrance area, thereby guiding vehicles to change lanes; When subsequent vehicles enter the tunnel entrance, the above steps are repeated to predict the travel time of each lane in the tunnel, thereby guiding the vehicle to change lanes.

2. The vehicle guidance method in a long tunnel based on radar-visual fusion data according to claim 1 is characterized by: In step S1, the data frame rate of the radar-visual fusion information is 10 frames per second.

3. The method for guiding vehicles in long tunnels based on radar-visual fusion data according to claim 2, characterized in that: In step S1, the traffic participant target data includes: ① vehicle id: Vehicle i ;②Lane number lanenum: lane i ③Speed: V i , unit km / h; ④ Position x: X i , unit: m.

4. The vehicle guidance method in a long tunnel based on radar-visual fusion data according to claim 3 is characterized by: The step S2 includes the following sub-steps: S2.1 sets the attenuation coefficient α and uses the first-order inertial filter to filter the vehicle i The trajectory data is smoothed every 0.1s; The filtering calculation formula is as follows: smooth_speed[i]=α*V[i]+(1-α)*smooth_speed[i-1] smooth_speed[0]=V[0] Where α represents the filter coefficient; smooth_speed[i] represents the output value of the current filter speed; smooth_speed[i-1] represents the output value of the previous filter speed; V[i] represents the current sampling speed value; V[0] represents the first frame speed; S2.2 removes the highest speed V among the 10 filtered speed output values ​​within 1s fast and the minimum speed V slow , using the median average method, the average value of the remaining 8 filtered speed output values ​​is obtained as Vehicle i Velocity information V at 1s intervals i .

5. The method for guiding vehicles in long tunnels based on radar and visual fusion data according to claim 4 is characterized by: The step S4 includes the following sub-steps: S4.1 Select Vehicle slow1 n sets of data before time t, including time t (t n ,t n-1 ,....t1) and the position x(x n ,x n-1 ,....x1), build the relationship between time and position: x=at+b Where x is the position, representing the dependent variable to be predicted; t is the time, representing the independent variable to be predicted; a is the slope of the regression line; b is a constant; S4.2 Solve for the values ​​of a and b: S4.3 Prediction at X slow1 The slowest vehicle slow1 Travel time to the tunnel exit T slow1 , the formula is as follows: Where Length represents the tunnel length.

Images