A method for determining an extreme lane-changing trajectory of an automatic driving vehicle at an intersection entrance

Abstract

The patent discloses a kind of intersection entrance automatic driving vehicle limit lane-changing trajectory determination method.The specific method mainly includes three steps: one, generates the lane-changing trajectory set based on quintic polynomial;Two, consider the optimization of lane-changing trajectory set to vehicle dynamics constraint;Three, traverse the lane-changing starting point position, determine limit lane-changing starting point and its corresponding limit lane-changing trajectory.The patent considers vehicle self condition constraint and road condition constraint, proposes the determination method of intersection entrance automatic driving vehicle limit lane-changing trajectory, provides the early warning of limit lane-changing trajectory for automatic driving vehicle, to improve driving safety, lane-changing success rate and traffic efficiency.

Description

Technical Field

[0001] This invention belongs to the field of autonomous vehicle trajectory planning, specifically relating to a method for determining the extreme lane-changing trajectory of an autonomous vehicle at an intersection entrance. Background Technology

[0002] Trajectory planning is a key and challenging aspect of autonomous driving technology, with lane changing being a fundamental driving behavior. Lane changing at intersection entrances is mandatory and frequent, requiring consideration of more complex factors than on regular road sections. Inadequate lane-changing trajectory planning can negatively impact traffic efficiency, lead to lane-changing failures, or even result in vehicles crossing lane lines, violating traffic rules and threatening driving safety. Providing autonomous vehicles with warnings of extreme lane-changing trajectories can prevent these situations. Therefore, determining the extreme lane-changing trajectories for autonomous vehicles at intersection entrances is crucial for improving driving safety, lane-changing success rates, and traffic efficiency.

[0003] Existing research on trajectory planning is extensive, but most focuses on obstacle avoidance and rarely addresses intersection approach scenarios. Invention patent 202111661902.6 determines the optimal obstacle avoidance lane-changing return point and a polynomial curve-based obstacle avoidance lane-changing trajectory by solving a multi-objective optimization problem concerning comfort, obstacle avoidance efficiency, and sideslip, based on road environment, obstacle information, and current vehicle speed. Invention patent 202111406002.7 first generates candidate trajectories based on road type, vehicle pose, and speed information, then constructs and calculates the cost function of the candidate trajectories, selecting the candidate trajectory with the lowest cost as the optimal trajectory. However, previous research on the extreme lane-changing trajectories of autonomous vehicles at intersection approach lanes is lacking. Lane-changing behavior at intersection approach lanes differs from that on regular road sections. To ensure driving safety, avoid traffic violations, and successfully change to the target lane to meet steering requirements, autonomous vehicles need warnings of extreme lane-changing trajectories to guide their lane-changing operations. Therefore, in the field of trajectory planning, methods for determining the extreme lane-changing trajectories of autonomous vehicles at intersection approach lanes are essential.

[0004] Given the above background, there is an urgent need to propose a method for determining the extreme lane-changing trajectory of autonomous vehicles at intersection entrances. This method should consider both vehicle and road constraints, and use a fifth-order polynomial to determine the extreme lane-changing trajectory to improve driving safety, lane-changing success rate, and traffic efficiency. However, a search has revealed no existing reports on methods for determining the extreme lane-changing trajectory of autonomous vehicles at intersection entrances. Summary of the Invention

[0005] 1. A method for determining the extreme lane-changing trajectory of an automated driving vehicle at an intersection approach, comprising the following steps:

[0006] Step 1: Generate a set of lane-changing trajectories based on a fifth-order polynomial:

[0007] Step 1.1 Establish the Frenet coordinate system:

[0008] The lane where the autonomous vehicle was located before changing lanes is defined as the original lane. The autonomous vehicle will be referred to as the vehicle. The center line of the original lane is used as the reference line, and the vehicle itself is used as the origin. A Frenet coordinate system is established with the tangent vector s and the normal vector d of the reference line as the coordinate axes. The direction of the tangent vector s of the reference line is defined as the longitudinal direction, and the direction of the normal vector d of the reference line is defined as the transverse direction.

[0009] Step 1.2: Establish and solve the lane-changing trajectory model based on a fifth-order polynomial:

[0010] Establish fifth-degree polynomial functions s(t) and d(t) of longitudinal and lateral displacements with respect to a certain time t during the lane change in the Frenet coordinate system:

[0011]

[0012] Where, p k q k It is the coefficient at time t, k = 0, 1, 2, 3, 4, 5;

[0013] Solve for the longitudinal displacement as a function of time t, and then differentiate s(t) twice to obtain the longitudinal velocity and longitudinal acceleration as functions of time t. and

[0014]

[0015] The starting time of the lane change is t=0, and the longitudinal position s(0) and longitudinal velocity of the vehicle at this time are... and longitudinal acceleration Substituting into the above equation, we get p0 = s(0), and

[0016] The fifth-degree polynomial function s(t) can be written in matrix form as follows:

[0017]

[0018] The longitudinal position s(T) of the vehicle at the end time t=T of the lane change is determined by using the endpoint of the guide lane line of the intersection approach lane away from the intersection as a reference, and the speed limit of the intersection approach lane is defined as v. limit The maximum acceleration of the vehicle is a max The longitudinal component v of the speed limit at the intersection approach lane limit,s and the longitudinal component a of the vehicle's maximum accelerationmax,s Let be the longitudinal velocity and longitudinal acceleration of the vehicle at the end of the lane change, respectively. Then, let be the longitudinal position s(T) and longitudinal velocity v of the vehicle at the end of the lane change. limit,s and longitudinal acceleration a max,s Substituting into the matrix above, we obtain p3, p4, and p5;

[0019] Solve for the lateral displacement as a function of time t, and then differentiate d(t) twice to obtain the lateral velocity and lateral acceleration as functions of time t. and

[0020]

[0021] The starting time of the lane change is t=0, and at this time the vehicle's lateral position d(0) and lateral velocity are... and lateral acceleration Substituting these values ​​into the above equation, we obtain q0 = d(0). and

[0022] The fifth-degree polynomial function d(t) can be written in matrix form as follows:

[0023]

[0024] Using the endpoint of the guide lane line at the intersection approach as a reference, determine the lateral position d(T) of the vehicle at the end of the lane change, and then define the lateral component v of the speed limit at the intersection approach. limit,d And the lateral component of the vehicle's maximum acceleration, a max,d Let be the lateral velocity and lateral acceleration of the vehicle at the end of the lane change, respectively. Then, let be the lateral position d(T) and lateral velocity v of the vehicle at the end of the lane change. limit,d and lateral acceleration a max,d Substituting these values ​​into the matrix above, we obtain q3, q4, and q5.

[0025] Finally, the lane change end time T is set to [3 seconds, 10 seconds] with a time step of 0.2 seconds to generate the lane change trajectory set;

[0026] Step 2: Optimize the lane-changing trajectory set considering vehicle dynamics constraints:

[0027] Incorporate vehicle dynamics constraints to optimize the lane-changing trajectory set generated in step one:

[0028] Speed ​​constraint: The speed v(t) of a vehicle during lane changing must not exceed the speed limit of the intersection approach lane, i.e., 0 ≤ v(t) ≤ v limit Acceleration constraint: The vehicle's acceleration a(t) during lane changing must not exceed the vehicle's maximum acceleration a. maxNot less than the vehicle's maximum deceleration a -max That is, a -max ≤a(t)≤a max Lateral acceleration constraints: During lane changing, the lateral acceleration of the vehicle should ensure that the vehicle does not skid or tilt, i.e. Where B is the vehicle track width, in cm or h. g Height of the center of mass, unit: cm, h r The height of the tilt center is in cm, and g is the acceleration due to gravity. Vehicle roll rate, unit: rad / g The road surface adhesion coefficient;

[0029] Step 3: Traverse the starting positions of lane changes to determine the extreme lane change starting points and their corresponding extreme lane change trajectories:

[0030] The area where the approach lane of an intersection can cross the lane divider line is called the lane-changing area. With a time step of 0.1 seconds, in the lane-changing area, the starting position of the lane-changing point is traversed from the vehicle position (s(0), d(0)) at the start time of lane changing in step one towards the intersection until only one track remains in the lane-changing trajectory set obtained in step two. The starting point of lane changing when the traversal stops is the extreme lane-changing point, and its corresponding trajectory is the extreme lane-changing trajectory.

[0031] Lane-changing operations that cross two or more lanes need to be calculated multiple times. Each time, the lane can only be changed to the adjacent lane. Steps one to three are repeated each time the extreme lane-changing starting point and extreme lane-changing trajectory are determined.

[0032] Compared with the prior art, the beneficial effects of the present invention are:

[0033] The present invention provides a method for determining the extreme lane-changing trajectory of an autonomous vehicle at an intersection entrance. This method considers both the vehicle's own constraints and road conditions, providing an early warning of the extreme lane-changing trajectory for the autonomous vehicle. This avoids violations of traffic rules, such as vehicles crossing lane lines during lane changes, and also improves driving safety, lane-changing success rate, and traffic efficiency. It provides technical support for the rational planning of lane-changing trajectories for autonomous vehicles. Attached Figure Description

[0034] Figure 1 This is the overall flowchart of the present invention;

[0035] Figure 2 This is a schematic diagram of the Frenet coordinate system of the present invention;

[0036] Figure 3 These are simulation calculation results for steps one to three of this invention;

[0037] Figure 4This is an example diagram of the extreme lane change trajectory of a single lane change according to the present invention;

[0038] Figure 5 This is an example diagram of the extreme lane-changing trajectory for multiple lane changes according to the present invention. Detailed Implementation

[0039] This invention patent describes a method for determining the extreme lane-changing trajectory of an automated driving vehicle at an intersection entrance. This method is applicable to determining an extreme lane-changing trajectory that ensures vehicle safety and compliance with traffic rules during lane-changing before entering an intersection, provided there is no interference from other vehicles. This extreme lane-changing trajectory does not intersect with guide lane lines, satisfies vehicle dynamics constraints, and the vehicle cannot perform a lane-changing operation after passing the extreme lane-changing starting point. Furthermore, vehicles that have just reached the extreme lane-changing starting point must follow the planned extreme lane-changing trajectory.

[0040] See Figure 1 The method for determining the extreme lane-changing trajectory of an automated driving vehicle at an intersection entrance, as described in this invention, comprises the following steps:

[0041] Step 1: In the Frenet coordinate system, as shown... Figure 2 As shown, a lane-changing trajectory model based on a fifth-order polynomial is established and solved to generate a set of lane-changing trajectories;

[0042] Step 2: Add vehicle dynamics constraints and optimize the lane-changing trajectory set generated in Step 1;

[0043] Step 3: Starting from the vehicle position at the beginning of the lane change in Step 1, traverse the lane change starting points towards the intersection until only one track remains in the lane change trajectory set of Step 2. This determines the extreme lane change starting point and its corresponding extreme lane change trajectory.

[0044] The simulation results of steps one to three are as follows: Figure 3 As shown, the extreme lane-changing trajectory for a single lane change determined by this invention is as follows: Figure 4 As shown, the endpoint of the extreme lane change is the position of the vehicle (s(T), d(T)) at the end of the lane change, which is determined with reference to the endpoint of the guide lane line of the intersection entrance lane in the direction away from the intersection.

[0045] The Road Traffic Safety Law of the People's Republic of China stipulates that vehicles may not change more than two lanes consecutively at once. Therefore, lane-changing operations that cross more than two lanes need to be calculated multiple times. Each time, the vehicle can only change to an adjacent lane. Steps one through three are repeated each time the starting point and trajectory of the extreme lane change are determined. Figure 5 As shown.

[0046] Example

[0047] This invention provides an embodiment of a method for determining the extreme lane-changing trajectory of an automated vehicle at an intersection approach, presenting the implementation process and verification results. However, the scope of protection of this invention is not limited to the following embodiment.

[0048] A simulation experiment is used as an example. The west entrance of the intersection of Yuhua Road and Youyi Street in Shijiazhuang City is selected. Each lane of this entrance is 3.25 meters wide, and the guide lane is 70 meters long, totaling 7 lanes: two dedicated left-turn lanes, four dedicated straight-ahead lanes, and one dedicated right-turn lane. Based on the method for determining the extreme lane-changing trajectory of an automated driving vehicle at an intersection entrance according to the present invention, the extreme lane-changing trajectory of the automated driving vehicle between each lane is calculated. The main lane-changing operations are shown in the table below, and the simulation calculation results are as follows. Figure 3 As shown.

[0049] Table 1 Main lane-changing operations between lanes

[0050]

[0051] Observation and statistical analysis of aerial video data of the intersection's approach lanes during the same period revealed that 6 vehicles crossed the lane lines to change lanes in the prohibited lane-changing area, i.e., the guide lane line area. After adopting the method for determining the extreme lane-changing trajectory of autonomous vehicles at intersection approach lanes described in this invention, no vehicles crossed the lane lines to change lanes in the simulation experiment, and the average travel time of vehicles performing lane-changing operations near the guide lane lines in the simulation experiment was reduced by 9.3% compared to the actual scenario. Therefore, it is evident that using this method to provide early warning of extreme lane-changing trajectories for autonomous vehicles ensures vehicle driving safety and improves lane-changing success rate and traffic efficiency.

Claims

1. A method for determining the extreme lane-changing trajectory of an automated driving vehicle at an intersection approach, comprising the following steps: Step 1: Generate a set of lane-changing trajectories based on a fifth-order polynomial: Step 1.1 Establish the Frenet coordinate system: The lane where the autonomous vehicle was located before changing lanes is defined as the original lane. The autonomous vehicle will be referred to as the vehicle. The center line of the original lane is used as the reference line, and the vehicle itself is used as the origin. A Frenet coordinate system is established with the tangent vector s and the normal vector d of the reference line as the coordinate axes. The direction of the tangent vector s of the reference line is defined as the longitudinal direction, and the direction of the normal vector d of the reference line is defined as the transverse direction. Step 1.2: Establish and solve the lane-changing trajectory model based on a fifth-order polynomial: Establish fifth-degree polynomial functions s(t) and d(t) of longitudinal and lateral displacements with respect to a certain time t during the lane change in the Frenet coordinate system: in, p k q k It is the coefficient at time t, k = 0, 1, 2, 3, 4, 5; Solve for the longitudinal displacement as a function of time t, and then differentiate s(t) twice to obtain the longitudinal velocity and longitudinal acceleration as functions of time t. and The starting time of the lane change is t=0, and the longitudinal position s(0) and longitudinal velocity of the vehicle at this time are... and longitudinal acceleration Substituting these values ​​into the above equation, we obtain p0 = s(0). and The fifth-degree polynomial function s(t) can be written in matrix form as follows: The longitudinal position s(T) of the vehicle at the end time t=T of the lane change is determined by using the endpoint of the guide lane line of the intersection approach lane away from the intersection as a reference, and the speed limit of the intersection approach lane is defined as v. limit The maximum acceleration of the vehicle is a max The longitudinal component v of the speed limit at the intersection approach lane limit,s and the longitudinal component of the vehicle's maximum acceleration a max,s Let be the longitudinal velocity and longitudinal acceleration of the vehicle at the end of the lane change, respectively. Then, let be the longitudinal position s(T) and longitudinal velocity v of the vehicle at the end of the lane change. limit,s and longitudinal acceleration a max,s Substituting into the matrix above, we obtain p3, p4, and p5; Solve for the lateral displacement as a function of time t, and then differentiate d(t) twice to obtain the lateral velocity and lateral acceleration as functions of time t. and The starting time of the lane change is t=0, and at this time the vehicle's lateral position d(0) and lateral velocity are... and lateral acceleration Substituting these values ​​into the above equation, we obtain q0 = d(0). and The fifth-degree polynomial function d(t) can be written in matrix form as follows: Using the endpoint of the guide lane line at the intersection approach as a reference, determine the lateral position d(T) of the vehicle at the end of the lane change, and then define the lateral component v of the speed limit at the intersection approach. limit,d And the lateral component of the vehicle's maximum acceleration, a max,d Let be the lateral velocity and lateral acceleration of the vehicle at the end of the lane change, respectively. Then, let be the lateral position d(T) and lateral velocity v of the vehicle at the end of the lane change. limit,d and lateral acceleration a max,d Substituting these values ​​into the matrix above, we obtain q3, q4, and q5. Finally, the lane change end time T is set to [3 seconds, 10 seconds] with a time step of 0.2 seconds to generate the lane change trajectory set; Step 2: Optimize the lane-changing trajectory set considering vehicle dynamics constraints: Incorporate vehicle dynamics constraints to optimize the lane-changing trajectory set generated in step one: Speed ​​constraint: The speed v(t) of a vehicle during lane changing must not exceed the speed limit of the intersection approach lane, i.e., 0 ≤ v(t) ≤ v limit Acceleration constraint: The vehicle's acceleration a(t) during lane changing must not exceed the vehicle's maximum acceleration a. max Not less than the vehicle's maximum deceleration a -max That is, a -max ≤a(t)≤a max Lateral acceleration constraints: During lane changing, the lateral acceleration of the vehicle should ensure that the vehicle does not skid or tilt, i.e. Where B is the vehicle track width, in cm or h. g Height of the center of mass, unit: cm, h r The height of the tilt center is in cm, and g is the acceleration due to gravity. Vehicle roll rate, unit: rad / g The road surface adhesion coefficient; Step 3: Traverse the starting positions of lane changes to determine the extreme lane change starting points and their corresponding extreme lane change trajectories: The area where the approach lane of an intersection can cross the lane divider line is called the lane-changing area. With a time step of 0.1 seconds, in the lane-changing area, the vehicle position (s(0), d(0)) at the start of the lane change in step one is traversed towards the intersection until only one track remains in the lane-changing trajectory set obtained in step two. The lane-changing start point when the traversal stops is the extreme lane-changing start point, and its corresponding trajectory is the extreme lane-changing trajectory. Lane-changing operations that cross two or more lanes need to be calculated multiple times. Each time, the lane can only be changed to the adjacent lane. Steps one to three are repeated each time the extreme lane-changing starting point and extreme lane-changing trajectory are determined.

Images