A left turn planning method, system, device and medium of an autonomous vehicle

Abstract

This invention provides a method, system, device, and medium for left-turn planning of autonomous vehicles, comprising: acquiring road traffic information, including the location information and driving information of all vehicles; constructing a vehicle safety field based on the location information; importing the driving information into the vehicle safety field to calculate the minimum safe distance; determining the risk distance between the autonomous vehicle and all other vehicles based on the driving information of all vehicles and the minimum safe distance; calculating the interaction risk value between the autonomous vehicle and all other vehicles based on the driving information and the risk distance; determining a control strategy based on the interaction risk value; and controlling the autonomous vehicle to make a left turn. This invention solves the problems of decision lag and difficulty in quantifying road risks in existing autonomous driving decision-making methods.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and in particular to a method, system, device and medium for left-turn planning of autonomous vehicles. Background Technology

[0002] In a connected and intelligent environment, turning left at a T-junction is a typical high-complexity, high-stakes scenario faced by autonomous vehicles. In mixed traffic flows where connected and manually driven vehicles coexist, the behavior patterns of traffic participants are complex, variable, and difficult to predict. Traditional autonomous driving decision-making methods generally rely on single-vehicle perception, which can only acquire information from a small area around the vehicle. This results in problems such as limited field of vision, blind spots, and data redundancy, which can easily lead to delayed or erroneous decisions.

[0003] In addition, existing autonomous driving decision-making systems typically use rule-based decision-making logic based on simple threshold judgments (such as distance, speed, and TTC), lacking the quantification of clear priority rules and the need to handle complex interactions among multiple traffic participants from multiple directions and types. This makes it difficult to quantify road risks and limits the accuracy and safety of decision-making. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method, system, device, and medium for left-turn planning of autonomous vehicles, which solves the problems of decision lag and difficulty in quantifying road risks in existing autonomous driving decision-making methods.

[0005] According to an embodiment of the present invention, a left-turn planning method for an autonomous vehicle includes: Obtain road traffic information, including the location information and driving information of all vehicles; construct a vehicle safety field based on the location information; then import the vehicle driving information into the vehicle safety field to calculate the minimum safe distance. Based on the vehicle driving information and minimum safe distance of all vehicles, determine the risk distance between your vehicle and all other vehicles; Calculate the interaction risk value between the vehicle and all other vehicles based on vehicle driving information and risk distance; The control strategy is determined based on the interaction risk value, and the vehicle is controlled to make a left turn.

[0006] Preferably, before constructing a safe field for vehicles, the speed of all vehicles is adjusted based on traffic perception data so that all vehicles travel at the optimal speed when they reach the intersection.

[0007] Preferably, the method for constructing a vehicle safety field based on road traffic information includes: Establish a local coordinate system with the vehicle's location as the origin and the direction of travel as the positive x-axis; Project the position information of all vehicles onto the projection coordinate system and calculate the difference in planar coordinates between your vehicle and other vehicles; The positions of other vehicles in the local coordinate system are determined based on the difference in planar coordinates, and then a vehicle safety field is constructed based on the positions of the vehicle itself and other vehicles.

[0008] Preferably, the vehicle safety field is as follows: in, , , , , All are constants. Let be the angle between the vehicle and other vehicles at time t. Let be the coordinates of the vehicle at time t. Let be the coordinates of other vehicles at time t. Let t be the speed of the vehicle. and The distance vectors between the vehicle and other vehicles in the x and y directions at time t, respectively.

[0009] Preferably, the method for calculating the risk distance between the vehicle and all other vehicles based on road traffic information includes: The vehicle driving information of the vehicle and other vehicles is extracted from the road traffic information. Then, based on the vehicle driving information, multiple predicted trajectories of the vehicle and other vehicles are generated using a trajectory prediction model. Based on the predicted trajectory of the vehicle and the predicted trajectories of other vehicles, the conflict trajectory points between the vehicle and other vehicles are determined, and then the risk distance between the vehicle and the trajectory conflict points is calculated.

[0010] Preferably, the method for determining the conflict trajectory points between the vehicle and other vehicles based on the vehicle's predicted trajectory and the predicted trajectories of other vehicles includes: Set the sampling interval, and then sample each predicted trajectory of the vehicle based on the sampling interval to obtain multiple corresponding trajectory points and corresponding coordinates. Then, combine all the trajectory points of the vehicle into a corresponding set of trajectory points. Based on the sampling interval, each other vehicle predicted trajectory is sampled to obtain multiple other vehicle trajectory points and their corresponding coordinates. Then, all other vehicle trajectory points are combined into a corresponding set of other vehicle trajectory points. Combine all the sets of trajectory points of the self-driving vehicle with the sets of trajectory points of all other vehicles to obtain multiple pairs of trajectory points; For each pair of trajectory points, calculate the first distance between the trajectory points of the vehicle and the trajectory points of other vehicles, and take all trajectory point pairs whose first distance is less than the minimum safety distance as candidate trajectory point pairs. Then, form a candidate point set from all candidate trajectory point pairs. The most frequent candidate trajectory point pair in the entire candidate point set is identified, and the midpoint between the two trajectory points in the candidate trajectory point pair is then taken as the trajectory conflict point.

[0011] Preferably, if the distance between all the trajectory points of the autonomous vehicle and all other vehicle trajectory points is greater than a preset value, then the distance between the autonomous vehicle and other vehicles is taken as the risk distance.

[0012] On the other hand, according to embodiments of the present invention, a left-turn planning system for an autonomous vehicle is also provided. This system uses the aforementioned left-turn planning method for an autonomous vehicle, comprising: Information acquisition module, the information acquisition module is used to acquire road traffic information; The data processing module is used to construct a vehicle safety field based on location information; The analysis module is used to import vehicle driving information into the vehicle safety field, calculate the minimum safe distance, determine the risk distance between the vehicle and all other vehicles based on the vehicle driving information and the minimum safe distance of all vehicles, and calculate the interaction risk value between the vehicle and all other vehicles based on the vehicle driving information and the risk distance. The control module is used to determine the control strategy based on the interaction risk value and control the vehicle to turn left.

[0013] On the other hand, according to an embodiment of the present invention, a computer is also provided, including at least one processor and a memory, the memory storing a computer program configured to be executed by the processor to implement the above-described method for planning left turns of an autonomous vehicle.

[0014] On the other hand, according to an embodiment of the present invention, a storage medium is also provided, which is a computer-readable storage medium, and a computer program is stored on the storage medium. The computer program can be executed by one or more processors to implement the above-described method for planning left turns of an autonomous vehicle.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This invention expands the information acquisition range of the autonomous vehicle by acquiring road traffic information containing information on all vehicles on the road. This allows for accurate and timely judgment of the driving status of surrounding vehicles and safe and accurate planning of the autonomous vehicle's subsequent driving strategy. Then, by constructing a vehicle safety field, the minimum safe distance between the autonomous vehicle and other vehicles is dynamically assessed. Finally, the road traffic information is quantified into the interaction risk value between the autonomous vehicle and other vehicles, providing specific judgment basis for the autonomous driving decision-making system and improving the accuracy of decision-making. Attached Figure Description

[0016] Figure 1 This is a diagram illustrating the left-turn planning method according to an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] like Figure 1 As shown in the figure, this embodiment of the invention proposes a left-turn planning method for autonomous vehicles, including: Obtain road traffic information, including the location information and driving information of all vehicles; construct a vehicle safety field based on the location information; then import the vehicle driving information into the vehicle safety field to calculate the minimum safe distance. A T-junction traffic scenario includes a main road and a side road perpendicular to the main road. The intersection of the main road and the side road is a core area where multiple vehicle trajectories intersect and there is a potential collision risk. In most scenarios, T-junctions are equipped with traffic lights to guide traffic. However, some areas do not have traffic lights or the traffic lights are broken. In this case, vehicles can pass freely, which will lead to a greater risk of collision in the intersection area, especially for autonomous vehicles.

[0019] In this invention, it is assumed that the vehicle is an autonomous vehicle traveling on the main road, intending to turn left into the side road. Then, the global perception capability of the roadside cooperative control unit (RCCU) is fully utilized to provide the vehicle with real-time traffic situation information that is beyond line of sight, has no blind spots, is highly accurate, and has low latency through V2X communication. This compensates for the blind spots of the vehicle's perception and accurately identifies and predicts the motion state and intentions of all traffic participants. After collecting road traffic information (including the location information and driving information of all vehicles) of all traffic participants (including the vehicle, other vehicles, and pedestrians) on the road, the roadside cooperative control unit sends it to the autonomous driving decision-making system of the vehicle. The decision-making system uses this road traffic information to make decisions and control the vehicle to turn left.

[0020] After acquiring road traffic information, the vehicle's speed is smoothly adjusted to ensure it reaches the entrance to the intersection area at optimal speed and comfort within a predetermined time window. This pre-coordinated speed significantly reduces the burden of complex trajectory adjustments and planning within the core traffic area, improving overall traffic efficiency and driving comfort. The guiding speed command is sent to the intelligent connected vehicle via V2X, and the intelligent connected vehicle's onboard unit performs simple speed tracking control.

[0021] Since the vehicle location information obtained by RUUC is in the world coordinate system of latitude and longitude, this invention needs to convert the latitude and longitude coordinates to the local coordinate system of the current traffic scenario. For ease of study, a local coordinate system is constructed with the vehicle's center of mass as the origin and the vehicle's direction of travel as the positive x-axis. The vehicle's latitude and longitude information is assumed to be... The heading angle is The latitude and longitude information of other vehicles is The heading angle is .

[0022] ① First, project the latitude and longitude information of all vehicles onto the projected coordinate system. Assume that the projected coordinates of your own vehicle and other vehicles are as follows: and .

[0023] ② The difference in plane coordinates is calculated as follows: ③ The relative coordinate system is calculated as follows: The driving risks during lane changes mainly come from oncoming lanes and side lanes. Meanwhile, the lateral and longitudinal velocity components generated by the vehicle cannot be ignored. Therefore, both lateral and longitudinal distance vectors must be considered. Thus, this invention constructs the following vehicle safety field: in, , , , , All are constants. Let be the angle between the vehicle and other vehicles at time t. Let be the coordinates of the vehicle at time t. Let be the coordinates of other vehicles at time t. Let t be the speed of the vehicle. and The distance vectors between the vehicle and other vehicles in the x and y directions at time t, respectively.

[0024] By improving the definition of the driving safety potential field, this method fully considers information such as vehicle speed, acceleration, and lane-changing intentions, and can accurately depict the driving risks and impact areas during the lane-changing process.

[0025] To ensure that the vehicle maintains a safe distance from vehicles in the oncoming lane and on the side lane when making a left turn, it is necessary to calculate the minimum safe distance between vehicles so that the distance between the vehicle and any other vehicle is always greater than the minimum safe distance.

[0026] Vehicle driving information mainly includes the vehicle's speed and acceleration. When the vehicle accelerates... ,speed , Take the longest braking distance (When the present invention takes 50 meters), it can be obtained that... The risks associated with lane changes primarily involve lateral collisions between the vehicle and vehicles in the adjacent lane or oncoming lanes. These risks vary over time. Since the underlying principles of these collisions are the same, they can be discussed together. Assuming all vehicles are of uniform specifications, then: in, and Let be the accelerations of the vehicle and the other vehicle at time t, respectively. and Let be the speeds of the vehicle and the other vehicle at time t, respectively.

[0027] Under the premise that the motion state of other vehicles remains unchanged, the time required for the speed of the vehicle to decrease to 0 is defined as the safety critical time. This safety critical time is the time to ensure the absolute safety of the vehicle. Therefore, the situation where the acceleration / deceleration of the vehicle causes the two vehicles to pass each other out of alignment is not considered.

[0028] Generally, if the two vehicles do not collide within this time period, the entire passage process can be guaranteed to be free of collision risk. Therefore, the safety critical time is calculated as follows: Based on the above discussion, to avoid collisions with other vehicles, a vehicle should meet the following requirements: in, For vehicle width, Let be the minimum safe distance at time t.

[0029] Based on the vehicle driving information and minimum safe distance of all vehicles, determine the risk distance between your vehicle and all other vehicles; After determining the minimum safe distance, it is necessary to judge whether a collision will occur between the vehicle and other vehicles, and to identify the point of conflict. Therefore, this invention is based on the current state (position) of the vehicle and other vehicles in the acquired road traffic information. and heading angle ,speed Add Gaussian noise perturbation, where the heading angle perturbation follows a normal distribution. ,in (Initial values, based on driving behavior uncertainty) Using existing trajectory prediction models (such as GAT and Transform deep learning models) to predict the trajectories of the vehicle and other vehicles respectively, multiple predicted trajectories of the vehicle and multiple predicted trajectories of other vehicles are generated.

[0030] Set time step Prediction window (Based on human reaction time and vehicle dynamic calibration), total number of steps For each time step : The vehicle's trajectory points can be extracted from the predicted trajectory of the vehicle: Trajectory points of other vehicles can be extracted from their predicted trajectories: in For the first The perturbation heading angle of the step.

[0031] All the vehicle trajectory points are combined into a corresponding set of vehicle trajectory points. All other vehicle trajectory points are combined into a corresponding set of vehicle trajectory points. Then, all the vehicle trajectory point sets are combined with all other vehicle trajectory point sets to obtain multiple sets of trajectory point pairs. All trajectory point pairs are traversed, and the first distance between the vehicle trajectory point and other vehicle trajectory points in each set of trajectory point pairs is calculated.

[0032] If the first distance Record the trajectory point pair as a candidate trajectory point pair, and then form a candidate point set from all candidate trajectory point pairs.

[0033] The pair of candidate trajectory points that appears most frequently in the entire candidate point set is identified, and the midpoint between the two trajectory points in the candidate trajectory is then taken as the trajectory conflict point. Then, the risk distance is calculated based on the coordinates of the vehicle's front center and the coordinates of the trajectory conflict point. If the distance between all the trajectory points of the self-vehicle and all the trajectory points of all other vehicles is greater than the minimum safe distance, then the distance between the self-vehicle and other vehicles is taken as the risk distance.

[0034] Calculate the interaction risk value between the vehicle and all other vehicles based on vehicle driving information and risk distance; The formula for calculating the interaction risk value between the vehicle and the trajectory conflict point is as follows: in, They are respectively in Time and heading angle The longitudinal speed of the vehicle and other vehicles. They are respectively in Time and heading angle The current longitudinal acceleration of the vehicle and other vehicles. For risk distance, Let t be the heading angular velocity of other vehicles at time t, reflecting their lateral movement trend or intention change.

[0035] The control strategy is determined based on the interaction risk value, and the vehicle is controlled to make a left turn.

[0036] Based on the interaction risk value, the present invention formulates the control strategies into the following categories: (1) High risk like If a vehicle is deemed to be in a high-risk state, it needs to slow down to a low speed and come to a complete stop while waiting for a safe opportunity.

[0037] At this point, the vehicle will employ a deceleration and stopping strategy until it comes to a complete stop or waits for a safe opportunity.

[0038] In the formula: The first target speed for the vehicle in the next moment. The vehicle's current speed. The distance between the current vehicle and the point of conflict with the trajectory. The distance required for a vehicle to park safely, or a safe distance, default value. , The velocity decay distance determines the gradient at which the velocity decreases; the default value is... .

[0039] This formula indicates that when the distance from the point of conflict with the trajectory... Much larger At that time, the speed was close to ;when near At that point, the speed will drop sharply and eventually come to a stop.

[0040] (2) Medium risk like The vehicle was assessed as being in a medium-risk state. The vehicle made a cautious left turn at a restricted speed, ready to stop at any time.

[0041] At this time, the maximum speed limit for the vehicle is as follows: In the formula, This is the maximum speed limit for the vehicle. This is the maximum permissible speed for a vehicle turning left; the default value is... , The current distance between the vehicle and other road users. The desired minimum TTC time is the time it would take for the vehicle and other road users to collide if the vehicle continues in its current state. (Default value) .

[0042] (3) Low risk like If the vehicle is deemed to be in a low-risk state, it can coordinate with V2X commands or HDV predicted intentions to make a left turn.

[0043] At this moment, the cooperative turning speed of the vehicle is: In the formula, For the vehicle to coordinate turning speed, For the speed of other vehicles, The default value represents the expected deceleration of other vehicles when they need to slow down. This represents the current distance between your vehicle and other vehicles.

[0044] (4) No risk like Once the vehicle is deemed to be in a risk-free state, it can freely turn left along the preset ideal trajectory.

[0045] The four decision states mentioned above use quantitative data that includes the interaction between the vehicle and other traffic participants, as well as the interaction between the vehicle and the environment. By integrating all traffic information, a comprehensive judgment can be made to enable efficient collaboration between the vehicle and other traffic participants, thereby further improving the decision-making accuracy of the autonomous driving decision-making system.

[0046] On the other hand, embodiments of the present invention also provide a left-turn planning system for autonomous vehicles, which uses the aforementioned left-turn planning method for autonomous vehicles, including: Information acquisition module, the information acquisition module is used to acquire road traffic information; The data processing module is used to construct a vehicle safety field based on location information; The analysis module is used to import vehicle driving information into the vehicle safety field, calculate the minimum safe distance, determine the risk distance between the vehicle and all other vehicles based on the vehicle driving information and the minimum safe distance of all vehicles, and calculate the interaction risk value between the vehicle and all other vehicles based on the vehicle driving information and the risk distance. The control module is used to determine the control strategy based on the interaction risk value and control the vehicle to turn left.

[0047] On the other hand, embodiments of the present invention also provide a computer, including at least one processor and a memory, the memory storing a computer program configured to be executed by the processor to implement the above-described method for planning left turns of an autonomous vehicle.

[0048] On the other hand, embodiments of the present invention also provide a storage medium, which is a computer-readable storage medium storing a computer program that can be executed by one or more processors to implement the above-described method for planning left turns of an autonomous vehicle.

[0049] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A left-turn planning method for an autonomous vehicle, characterized in that: include: Obtain road traffic information, including the location information and driving information of all vehicles; construct a vehicle safety field based on the location information; then import the vehicle driving information into the vehicle safety field to calculate the minimum safe distance. Based on the vehicle driving information and minimum safe distance of all vehicles, determine the risk distance between your vehicle and all other vehicles; Calculate the interaction risk value between the vehicle and all other vehicles based on vehicle driving information and risk distance; The control strategy is determined based on the interaction risk value, and the vehicle is controlled to make a left turn.

2. The left-turn planning method for an autonomous vehicle as described in claim 1, characterized in that: Before constructing a safe field for vehicles, speed control is applied to all vehicles based on traffic perception data to ensure that all vehicles travel at the optimal speed when approaching the intersection.

3. The left-turn planning method for an autonomous vehicle as described in claim 1, characterized in that: Methods for constructing vehicle safety fields based on road traffic information include: Establish a local coordinate system with the vehicle's location as the origin and the direction of travel as the positive x-axis; Project the position information of all vehicles onto the projection coordinate system and calculate the difference in planar coordinates between your vehicle and other vehicles; The positions of other vehicles in the local coordinate system are determined based on the difference in planar coordinates, and then a vehicle safety field is constructed based on the positions of the vehicle itself and other vehicles.

4. The left-turn planning method for an autonomous vehicle as described in claim 3, characterized in that: The vehicle safety field is as follows: in, , , , , All are constants. Let be the angle between the vehicle and other vehicles at time t. Let be the coordinates of the vehicle at time t. Let be the coordinates of other vehicles at time t. Let t be the speed of the vehicle. and The distance vectors between the vehicle and other vehicles in the x and y directions at time t, respectively.

5. The left-turn planning method for an autonomous vehicle as described in claim 1, characterized in that: Methods for calculating the risk distance between your vehicle and all other vehicles based on road traffic information include: The vehicle driving information of the vehicle and other vehicles is extracted from the road traffic information. Then, based on the vehicle driving information, multiple predicted trajectories of the vehicle and other vehicles are generated using a trajectory prediction model. Based on the predicted trajectory of the vehicle and the predicted trajectories of other vehicles, the conflict trajectory points between the vehicle and other vehicles are determined, and then the risk distance between the vehicle and the trajectory conflict points is calculated.

6. The left-turn planning method for an autonomous vehicle as described in claim 5, characterized in that: Methods for determining conflict trajectory points between a vehicle and other vehicles based on the vehicle's predicted trajectory and the predicted trajectories of other vehicles include: Set the sampling interval, and then sample each predicted trajectory of the vehicle based on the sampling interval to obtain multiple corresponding trajectory points and corresponding coordinates. Then, combine all the trajectory points of the vehicle into a corresponding set of trajectory points. Based on the sampling interval, each other vehicle predicted trajectory is sampled to obtain multiple other vehicle trajectory points and their corresponding coordinates. Then, all other vehicle trajectory points are combined into a corresponding set of other vehicle trajectory points. Combine all the sets of trajectory points of the self-driving vehicle with the sets of trajectory points of all other vehicles to obtain multiple pairs of trajectory points; For each pair of trajectory points, calculate the first distance between the trajectory points of the vehicle and the trajectory points of other vehicles, and take all trajectory point pairs whose first distance is less than the minimum safety distance as candidate trajectory point pairs. Then, form a candidate point set from all candidate trajectory point pairs. The most frequent candidate trajectory point pair in the entire candidate point set is identified, and the midpoint between the two trajectory points in the candidate trajectory point pair is then taken as the trajectory conflict point.

7. The left-turn planning method for an autonomous vehicle as described in claim 6, characterized in that: If the distance between all the trajectory points of the autonomous vehicle and all other vehicle trajectory points is greater than the preset value, then the distance between the autonomous vehicle and other vehicles will be used as the risk distance.

8. A left-turn planning system for an autonomous vehicle, characterized in that: The system uses a left-turn planning method for an autonomous vehicle as described in any one of claims 1-7, comprising: Information acquisition module, the information acquisition module is used to acquire road traffic information; The data processing module is used to construct a vehicle safety field based on location information; The analysis module is used to import vehicle driving information into the vehicle safety field, calculate the minimum safe distance, determine the risk distance between the vehicle and all other vehicles based on the vehicle driving information and the minimum safe distance of all vehicles, and calculate the interaction risk value between the vehicle and all other vehicles based on the vehicle driving information and the risk distance. The control module is used to determine the control strategy based on the interaction risk value and control the vehicle to turn left.

9. A computer, characterized in that: It includes at least one processor and a memory, the memory storing a computer program configured to be executed by the processor to implement a left-turn planning method for an autonomous vehicle according to any one of claims 1-7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, on which a computer program is stored. The computer program can be executed by one or more processors to implement a left-turn planning method for an autonomous vehicle as described in any one of claims 1-7.

Images