Unprotected left turn method for intelligent network connection motorcade

Abstract

The invention relates to an unprotected left turn method for an intelligent network connection fleet, and the method comprises the following steps: 1, determining conflicting vehicles in opposite straight traffic flow when the fleet enters an intersection and starts to turn left; 2, according to the historical operation data of the conflicting vehicles, the driving behaviors of the conflicting vehicles are identified, and a conflicting vehicle driving behavior model is obtained; 3, building a model prediction control problem according to the real-time state of the motorcade and the driving behavior model of the conflicting vehicles, and solving to obtain a motorcade left-turn track; and 4, driving each vehicle in the vehicle team according to the left-turn track of the vehicle team. Compared with the prior art, the method has the advantages of improving the safety and efficiency of left turning of the motorcade and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent networked vehicles, and in particular to an unprotected left turn method for an intelligent networked vehicle fleet. Background Art

[0002] With the development of intelligent and connected vehicles, intelligent connected fleets have become a new technology. Through the cooperation between vehicles, intelligent connected fleets can effectively shorten the distance between vehicles, thus helping to reduce vehicle energy consumption and improve traffic operation efficiency.

[0003] However, at present, the research on intelligent networked fleets rarely considers the direct conflicts with surrounding vehicles. Especially in the scenario of unprotected left turn at intersections, there are conflicts and bargaining between intelligent networked fleets and oncoming straight traffic, which seriously affects the safety and efficiency of intelligent networked fleets. Summary of the invention

[0004] The purpose of the present invention is to improve the safety of left turns of a fleet, make the trajectory planning of the fleet more proactive and improve the efficiency of left turns while ensuring safety, and provide an unprotected left turn method for intelligent connected fleets.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] An unprotected left turn method for an intelligent connected vehicle fleet, the method comprising the following steps:

[0007] Step 1: When the convoy enters the intersection and starts to turn left, it identifies the conflicting vehicles in the oncoming straight traffic flow;

[0008] Step 2: Identify the driving behavior of the conflicting vehicle based on the historical operation data of the conflicting vehicle and obtain the driving behavior model of the conflicting vehicle;

[0009] Step 3: Based on the real-time status of the convoy and the driving behavior model of the conflicting vehicles, a model predictive control problem is constructed to solve the convoy's left-turn trajectory;

[0010] Step 4: Each vehicle in the convoy drives along the convoy's left-turn trajectory.

[0011] Furthermore, step 1 is specifically as follows:

[0012] Step 1.1: Sort the opposite through traffic flows according to their distance from the intersection, from near to far;

[0013] Step 1.2: Select the vehicles in the straight-moving traffic flow that meet the following conditions as conflicting vehicles:

[0014] The location does not exceed the conflict zone;

[0015] The distance between the vehicle and the conflict zone is greater than the distance between the convoy and the conflict zone;

[0016] Under uniform motion, the time it takes for a vehicle to reach the conflict zone is greater than the time it takes for the convoy to reach the conflict zone.

[0017] Furthermore, step 2 is specifically as follows:

[0018] Step 2.1: Give the conflict vehicle driving behavior model formula;

[0019] Step 2.2: Obtain historical operation data of conflicting vehicles before entering the intersection;

[0020] Step 2.3: Based on the data at each time point in the historical operation data, the least squares method is used to fit the conflicting vehicle driving behavior model by adjusting the parameters to be identified, and the optimal parameters are obtained to obtain the conflicting vehicle driving behavior model under the optimal parameters.

[0021] Furthermore, the conflicting vehicle driving behavior model formula is:

[0022] a CV =k 1 (k 2 +tanh(aΔx-b)-v CV )+k 3 (v front -v CV )

[0023] Among them, k j∈{1,2,3} is the parameter to be identified, Δx is the distance between the conflicting vehicle and the preceding vehicle; v front is the speed of the front vehicle; v CV is the speed of the conflicting vehicle; a CV is the acceleration of the conflicting vehicle; a and b are the other parameters, and the front vehicle is the vehicle projected in front of the conflicting vehicle by the first vehicle in the convoy.

[0024] Furthermore, step 3 is specifically as follows:

[0025] Step 3.1: Define the Frenet coordinate system;

[0026] Step 3.2: Define the state quantity of the fleet;

[0027] Step 3.3: Define the control quantity of the fleet and define the state update equation of the fleet based on the linearized bicycle model;

[0028] Step 3.4: Define the state update equation of the conflicting vehicles;

[0029] Step 3.5: Calculate the acceleration of the conflicting vehicle in the prediction time domain in the state update equation based on the conflicting vehicle driving behavior model;

[0030] Step 3.6: Set the objective function of the model predictive control problem based on the state update equation;

[0031] Step 3.7: Set the control quantity constraints of the state quantity;

[0032] Step 3.8: Set obstacle avoidance constraints;

[0033] Step 3.9: Solve the model predictive control problem consisting of the objective function, control quantity constraints and obstacle avoidance constraints.

[0034] Furthermore, the state update equation of the fleet is:

[0035]

[0036]

[0037] Among them, the subscript i represents the i-th prediction time domain; Δt represents the linearization step; l represents the wheelbase of the vehicle, the superscript k represents the k-th vehicle, and x, y, v, and ψ represent the x-axis position, y-axis position, speed, and heading angle, respectively.

[0038] Furthermore, the state update equation of the conflicting vehicles is:

[0039]

[0040] Among them, the superscript CV represents the conflicting vehicle, i represents the i-th prediction time domain, represents the acceleration of the conflicting vehicle in the prediction time domain i;

[0041] The acceleration of the conflicting vehicle in the prediction time domain i is:

[0042]

[0043] in, is the speed of the first car.

[0044] Furthermore, the objective function is:

[0045] J=J 1 +J 2

[0046]

[0047] Among them, v des is the desired speed of the fleet, gap des is the expected distance between vehicles in the convoy, a 1 ,a 2 ,a 3 ,a 4 ,a 5 ,a 6 ,a7 is the weighting factor, δ f Indicates the front wheel turning angle.

[0048] Furthermore, the control quantity constraint is that the speed of the vehicles in the convoy in the control quantity is less than the road speed limit and the acceleration and the front wheel turning angle in the control quantity meet the performance requirements of the vehicle.

[0049] Furthermore, the obstacle avoidance constraint is:

[0050]

[0051] Among them, x con is the location of the conflict zone; length is the vehicle length; l safe For extra safety distance.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] The present invention considers the future reaction of the conflicting vehicle in the modeling, that is, the reaction of the conflicting vehicle to the "convoy behavior". The acceleration of the conflicting vehicle is related to the distance between the conflicting vehicle and the front vehicle, that is, the first vehicle of the convoy after projection, and the speed of the front vehicle. Compared with the conventional "predict first, then plan" trajectory prediction method, the present invention improves the safety of the convoy's left turn. While ensuring safety, the trajectory planning of the convoy can be made more proactive, improving the efficiency of left turns. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 Determining a map for conflicting vehicles of the present invention;

[0055] Figure 2 A driving behavior diagram of the conflicting vehicles of the present invention;

[0056] Figure 3 It is the Frenet coordinate system definition diagram of the present invention;

[0057] Figure 4 It is the projection diagram of the fleet of the present invention. DETAILED DESCRIPTION

[0058] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0059] In the unprotected left turn scenario at the intersection, there are conflicts and bargaining between the intelligent connected vehicle fleet and the oncoming straight traffic, which affects the traffic safety and efficiency of the intelligent connected vehicle fleet. To solve this problem, the present invention proposes an unprotected left turn method for intelligent connected vehicle fleets. The method first identifies the driving behavior of the conflicting vehicle based on its historical operating data to obtain a driving behavior model of the conflicting vehicle. The model is applied to the left turn trajectory planning of the fleet, and the left turn trajectory of each vehicle in the fleet is planned based on the optimal control method. Compared with other methods, the present invention expands the control object from a single vehicle to a fleet, and improves the safety and efficiency of the fleet's left turn. At the same time, the algorithm has the computing efficiency of an engineering application level, which can significantly reduce the computing load.

[0060] The present invention comprises the following steps:

[0061] Step 1: When the convoy enters the intersection and starts to turn left, it identifies the conflicting vehicles in the oncoming straight traffic flow;

[0062] Step 2: Identify the driving behavior of the conflicting vehicle based on the historical operation data of the conflicting vehicle and obtain the driving behavior model of the conflicting vehicle;

[0063] Step 3: Based on the real-time status of the convoy and the driving behavior model of the conflicting vehicles, a model predictive control problem is constructed to solve the convoy's left-turn trajectory;

[0064] Step 4: Each vehicle in the convoy drives along the convoy's left-turn trajectory.

[0065] In step 1, if Figure 1 As shown, the method for determining the conflicting vehicles is as follows:

[0066] Step 1.1: Sort the opposite through traffic flows according to their distance from the intersection, from near to far.

[0067] Step 1.2: Select vehicles that meet the following conditions at the same time as candidate vehicles.

[0068] 1. The vehicle’s position does not exceed the conflict zone.

[0069] 2. Compared with the distance between the convoy and the conflict zone, this vehicle is farther away from the conflict zone.

[0070] 3. Assuming that both vehicles are moving at a constant speed, the vehicle takes longer to reach the conflict zone than the convoy.

[0071] Step 1.3: Select the vehicle with the smallest serial number as the conflicting vehicle.

[0072] In step 2, the driving behavior identification method of the conflicting vehicles is as follows:

[0073] Step 2.1: Given a driving behavior model, where the parameter k j∈{1,2,3} is the parameter to be identified.

[0074] a CV =k 1 (k 2 +tanh(aΔx-b)-v CV )+k 3 (v front -v CV )

[0075] Where Δx is the distance between the conflicting vehicle and the preceding vehicle, see Figure 2 ;v front is the speed of the front vehicle; v CV is the speed of the conflicting vehicle; a CV is the acceleration of the collision vehicle; a and b are other parameters.

[0076] Step 2.2: Get the trajectory data of the conflicting vehicles before they enter the intersection, as follows:

[0077]

[0078] The subscripts represent the time; T represents the current time. The state of the conflicting vehicle, the state of the leading vehicle, and the acceleration response of the conflicting vehicle in the [0, T] time period are recorded.

[0079] Step 2.3: Based on The data at each time point in the least square method is used to adjust the parameter k j∈{1,2,3} Fit the formula in step 2.1. Then get the optimal parameters

[0080] In step 3, the left-turn trajectory method of the convoy based on model predictive control includes the following steps. Without losing generality, a convoy of three vehicles is taken as an example below.

[0081] Step 3.1: Define the Frenet coordinate system, such as Figure 3 shown.

[0082] Step 3.2: Define the state of the fleet as follows.

[0083]

[0084] The superscripts 1, 2, and 3 represent the serial numbers of the vehicles in the fleet. 1 ,y 1 ,v 1 ,ψ 1 Respectively represent the x-axis position, y-axis position, speed, and heading angle of the first vehicle. CV ,v CV Indicates the y-axis position and speed of the conflicting vehicle.

[0085] Step 3.3: Define the control quantity of the fleet as follows.

[0086]

[0087] The superscripts 1, 2, and 3 represent the serial numbers of the vehicles in the fleet. 1 ,δ f 1 represent the acceleration and front wheel angle of the first car respectively.

[0088] Step 3.3: Define the state update equation of the fleet based on the linearized bicycle model. Taking the kth vehicle as an example, it is as follows

[0089]

[0090] Wherein, the subscript i represents the i-th prediction time domain; Δt represents the linearization step size; and l represents the wheelbase of the vehicle.

[0091] Step 3.4: Define the state update equations for the conflicting vehicles.

[0092]

[0093] in represents the acceleration of the conflicting vehicle at step i, which can be determined by step 3.5.

[0094] Step 3.5: The acceleration of the conflicting vehicle in the prediction time domain can be determined by the following formula:

[0095]

[0096] To determine Need to get Δx i and However, since the convoy and the conflicting vehicle are not in the same lane, the first vehicle in the convoy is projected in front of the conflicting vehicle. Figure 4 After projection, considering the geometric relationship, we can get Δx i . is the speed of the first car, that is

[0097] Step 3.6: Considering the safety and efficiency of the convoy's left turn, the objective function of the model predictive control problem is set as follows:

[0098] J=J 1 +J 2

[0099]

[0100]

[0101] where v des is the desired speed of the fleet, gap des is the expected distance between vehicles in the convoy. 1 The first item ensures driving efficiency, the second item improves safety by keeping a safe distance between the convoy and conflicting vehicles, the third item ensures that the convoy will not deviate seriously from the center line of the road, and the fourth and fifth items ensure the stability of the distance between vehicles within the convoy. 2 The first and second terms of can reduce the change of vehicle state. Weighting factor a 1 ,a 2 ,a 3 ,a 4 ,a 5 ,a 6 ,a 7 Can be preset and adjusted according to user needs.

[0102] Step 3.7: The model predictive control problem needs to consider the control quantity constraints of the state quantity, that is, the speed of the vehicles in the convoy must be less than the road speed limit; the acceleration and front wheel steering angle in the control quantity must meet the performance requirements of the vehicle.

[0103] Step 3.8: The model predictive control problem needs to consider the obstacle avoidance constraint as follows:

[0104]

[0105] where x con is the location of the conflict zone; length is the vehicle length; l safe For extra safety distance.

[0106] Step 3.9: Convert the problem into a mixed integer quadratic programming problem and solve it.

[0107] The present invention proposes an unprotected left turn method for intelligent networked fleets. The method first identifies the driving behavior of conflicting vehicles based on their historical operating data to obtain a driving behavior model of conflicting vehicles. The model is then applied to the left turn trajectory planning of the fleet, and the left turn trajectory of each vehicle in the fleet is planned based on the optimal control method.

[0108] The beneficial effects of the present invention are:

[0109] 1. An unprotected left-turn method for intelligent connected vehicle fleets is proposed, filling the gap in this research.

[0110] 2. The future reactions of conflicting vehicles are taken into account in the modeling, which improves the safety of the convoy's left turn. While ensuring safety, the trajectory planning of the convoy can be more proactive, improving the efficiency of left turns.

[0111] 3. The algorithm has the computational efficiency of engineering application level, reducing the computing load.

[0112] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. An unprotected left turn method for an intelligent connected vehicle fleet, characterized in that: The method comprises the following steps: Step 1: When the convoy enters the intersection and starts to turn left, it identifies the conflicting vehicles in the oncoming straight traffic flow; Step 2: Identify the driving behavior of the conflicting vehicle based on the historical operation data of the conflicting vehicle and obtain the driving behavior model of the conflicting vehicle; Step 3: Based on the real-time status of the convoy and the driving behavior model of the conflicting vehicles, a model predictive control problem is constructed to solve the convoy's left-turn trajectory; Step 4: Each vehicle in the convoy drives along the convoy's left-turn trajectory.

2. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 1, characterized in that: Step 1 is as follows: Step 1.1: Sort the opposite through traffic flows according to their distance from the intersection, from near to far; Step 1.2: Select the vehicles in the straight-moving traffic flow that meet the following conditions as conflicting vehicles: The location does not exceed the conflict zone; The distance between the vehicle and the conflict zone is greater than the distance between the convoy and the conflict zone; Under uniform motion, the time it takes for a vehicle to reach the conflict zone is greater than the time it takes for the convoy to reach the conflict zone.

3. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 2, characterized in that: Step 2 is as follows: Step 2.1: Give the conflict vehicle driving behavior model formula; Step 2.2: Obtain historical operation data of conflicting vehicles before entering the intersection; Step 2.3: Based on the data at each time point in the historical operation data, the least squares method is used to fit the conflicting vehicle driving behavior model by adjusting the parameters to be identified, and the optimal parameters are obtained to obtain the conflicting vehicle driving behavior model under the optimal parameters.

4. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 3, characterized in that: The conflict vehicle driving behavior model formula is: a CV =k1(k2+tanh(aΔx-b)-v CV )+k3(v front -v CV ) Among them, k j∈{1,2,3} is the parameter to be identified, Δx is the distance between the conflicting vehicle and the preceding vehicle; v front is the speed of the front vehicle; v CV is the speed of the conflicting vehicle; a CV is the acceleration of the conflicting vehicle; a and b are the other parameters, and the front vehicle is the vehicle projected in front of the conflicting vehicle by the first vehicle in the convoy.

5. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 1, characterized in that: Step 3 is as follows: Step 3.1: Define the Frenet coordinate system; Step 3.2: Define the state quantity of the fleet; Step 3.3: Define the control quantity of the fleet and define the state update equation of the fleet based on the linearized bicycle model; Step 3.4: Define the state update equation of the conflicting vehicles; Step 3.5: Calculate the acceleration of the conflicting vehicle in the prediction time domain in the state update equation based on the conflicting vehicle driving behavior model; Step 3.6: Set the objective function of the model predictive control problem based on the state update equation; Step 3.7: Set the control quantity constraints of the state quantity; Step 3.8: Set obstacle avoidance constraints; Step 3.9: Solve the model predictive control problem consisting of the objective function, control quantity constraints and obstacle avoidance constraints.

6. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 5, characterized in that: The state update equation of the fleet is: Among them, the subscript i represents the i-th prediction time domain; Δt represents the linearization step; l represents the wheelbase of the vehicle, the superscript k represents the k-th vehicle, and x, y, v, and ψ represent the x-axis position, y-axis position, speed, and heading angle, respectively.

7. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 6, characterized in that: The state update equation of the conflicting vehicles is: Among them, the superscript CV represents the conflicting vehicle, i represents the i-th prediction time domain, represents the acceleration of the conflicting vehicle in the prediction time domain i; The acceleration of the conflicting vehicle in the prediction time domain i is: in, is the speed of the first car.

8. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 7, characterized in that: The objective function is: J=J1+J2 Among them, v des is the desired speed of the fleet, gap des is the expected distance between vehicles in the convoy, a1, a2, a3, a4, a5, a6, a7 are weighting factors, δ f Indicates the front wheel turning angle.

9. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 8, characterized in that: The control quantity constraints are: the speed of the vehicles in the convoy in the control quantity is less than the road speed limit and the acceleration and front wheel steering angle in the control quantity meet the performance requirements of the vehicle.

10. The unprotected left turn method for an intelligent networked vehicle fleet according to claim 9, characterized in that: The obstacle avoidance constraints are: Among them, x con is the location of the conflict zone; length is the vehicle length; l safe For extra safety distance.

Images