A trajectory planning method for connected autonomous driving vehicles based on artificial potential field method

Abstract

This invention proposes a trajectory planning method for connected autonomous vehicles based on an artificial potential field approach, comprising the following steps: S1: constructing an artificial potential field based on the basic conditions and traffic conditions of the expressway, including a target point attraction potential field generated by the position and velocity of the target point; a dynamic environment vehicle repulsion potential field generated by the position and velocity of the dynamic environment vehicle; and a road boundary repulsion potential field generated by the road boundary; S2: generating an initial vehicle trajectory based on the artificial potential field by calculating the resultant force of the controlled vehicle in the potential field, and smoothing the initial trajectory; S3: feeding back the smoothed controlled vehicle state information to a controller, combining it with real-time data from surrounding vehicles to form a closed-loop control loop and dynamically adjust the vehicle trajectory. This method effectively improves traffic operation efficiency, reduces vehicle delays, and has a positive impact on reducing carbon emissions.

Description

Technical Field

[0001] The present invention relates to the field of intelligent traffic control technology, and in particular to a trajectory planning method for a networked autonomous driving vehicle based on an artificial potential field method. Background Art

[0002] With the rapid development of my country's economy and technology, people have higher demands for travel convenience, efficiency, and safety. With the rapid increase in the number of vehicles, traffic congestion and accidents are becoming increasingly serious. To address these issues, expressways have become a key infrastructure for alleviating traffic pressure in large cities. While expressways can alleviate some traffic pressure, they still experience regular traffic congestion, frequent accidents, and environmental pollution during peak hours due to insufficient traffic capacity in intersections.

[0003] The development of intelligent connected technology has revolutionized traffic control. Through intelligent information exchange and sharing between vehicles and between vehicles and infrastructure, more refined and intelligent traffic management can be achieved. Against this backdrop, the research and development of intelligent connected traffic control systems has become a hot topic and a key focus in the field of transportation engineering. It has also provided new solutions to traffic control issues in freeway weaving zones. Trajectory-level coordinated traffic control is an effective way to alleviate traffic congestion in weaving zones, improve road safety, and reduce environmental pollution.

[0004] However, existing intelligent connected traffic control systems face numerous challenges, including: 1) Insufficient research on trajectory-level traffic coordination. Current research focuses primarily on macroscopic vehicle behavior, such as flow control and congestion management, while relatively little in-depth study of the specific driving trajectories and spatiotemporal coordination of individual vehicles has been conducted. 2) Limitations of traditional lane divisions on autonomous vehicles: The integration of autonomous vehicles requires a rethinking of traditional lane divisions and traffic rules. Lanes may no longer be fixed but may instead dynamically adjust based on traffic conditions, thereby improving road flexibility and efficiency. 3) The diversity of traffic scenarios and driving behaviors: Existing traffic control research is mostly based on specific traffic scenarios or only considers a single dimension for trajectory planning of connected autonomous vehicles. 4) Limited dynamic application of artificial potential field methods: In dynamic environments, the influence of speed must be considered, and the movement of obstacles complicates path planning. Artificial potential field methods need to be improved to enable them to respond to environmental changes in real time and provide safe and efficient paths for vehicles. Summary of the Invention

[0005] The purpose of the present invention is to provide a trajectory planning method for intelligent connected autonomous driving vehicles that can take into account the impact of multiple factors on vehicle driving, so as to improve traffic operation efficiency in expressway weaving areas, reduce delays and achieve environmental protection goals.

[0006] To achieve the above objectives, the present invention proposes a trajectory planning method for a connected autonomous driving vehicle based on an artificial potential field method, comprising the following steps:

[0007] S1: Construct an artificial potential field based on the basic conditions and traffic conditions of the expressway, including:

[0008] The gravitational potential field of the target point generated by the position and velocity of the target point;

[0009] Dynamic environment vehicle repulsive potential field generated by dynamic environment vehicle position and velocity;

[0010] Road boundary repulsive potential field generated by the road boundary;

[0011] S2: generating an initial trajectory of the vehicle by calculating the resultant force (force condition) of the controlled vehicle in the artificial potential field, and smoothing the initial trajectory to reduce trajectory fluctuations;

[0012] S3: The controlled vehicle state after smoothing The information is fed back to the controller, which forms a closed-loop control based on the real-time data of the surrounding vehicles and dynamically adjusts the vehicle trajectory. is the center of mass position of the controlled vehicle; is the controlled vehicle speed; a c It represents the acceleration of the controlled vehicle and is determined by the resultant force F / m of the controlled vehicle. m is the mass of a standard car, which can be 1.5t.

[0013] Furthermore, the basic conditions of the expressway include the construction of the expressway's road boundary and the location of static obstacles on the expressway, and the traffic conditions include the real-time location and speed of surrounding vehicles. The application scenario of the present invention is a fully connected autonomous driving scenario, where the road structure is only width and has no road divisions. Lanes are not fixed, so vehicles can move freely in the horizontal and vertical two-dimensional space without being affected by lane divisions.

[0014] Furthermore, in step S1, the basic conditions and traffic conditions of the expressway include the gravitational potential field generated by the target point position and speed, the repulsive potential field generated by the dynamic environment vehicle position and speed, and the repulsive potential field generated by the road boundary.

[0015] Furthermore, in step S1, the method for constructing the target point gravitational potential field is: based on the relative position difference between the controlled vehicle and the target node and relative speed difference The target point attraction potential field can be regarded as the result of the merging area point moving along the target lane at the restricted speed. The target point attraction potential field function is defined as: ;

[0016] in, is the center of mass position of the controlled vehicle; is the target node position; is the controlled vehicle speed;

[0017] The target node speed is the speed limit of the expressway;

[0018] are the Euclidean distances of relative position and relative velocity, respectively, and their expressions are:

[0019] ;

[0020] ;

[0021] In the expression, 、 are the gain coefficients of target point position and velocity gravity respectively.

[0022] Furthermore, the gravitational function corresponding to the gravitational potential field function of the target point is the direction in which the potential field decreases fastest, that is, ;in, Used to make the controlled vehicle track the position of the target node, The speed used to make the controlled vehicle track the target node, is the gravitational function corresponding to the gravitational potential field function of the target point, which represents the gravitational potential field function The fastest falling direction and the maximum rate of change at that point; Represents the negative gradient of the gravitational potential field.

[0023] Furthermore, in step S1, the method for constructing the dynamic environment vehicle repulsive potential field is: based on the relative position of the controlled vehicle and the i-th environment vehicle and relative speed , determine the dynamic environment vehicle repulsive potential field function as:

[0024] ;

[0025] ;

[0026] in, 、 is the position and speed of the i-th environment vehicle, 、 are the environmental vehicle position and velocity repulsion gain coefficients respectively, is the maximum range of the environmental vehicle;

[0027] The repulsive potential field generated by the i-th environmental vehicle on the controlled vehicle;

[0028] ):Right now , represents the vector difference between the position of the controlled vehicle and the position of the i-th vehicle;

[0029] :Right now , indicating that the speed of the i-th vehicle is a vector difference with the speed of the i-th vehicle.

[0030] Furthermore, in the dynamic environment vehicle repulsive potential field function:

[0031] when ;

[0032] when hour, ;

[0033] when hour, ;

[0034] when hour, ;

[0035] The repulsive force function corresponding to the dynamic environment vehicle repulsive force potential field function is:

[0036] ;

[0037] in, : The repulsive potential field generated by the velocity of the i-th environmental vehicle on the controlled vehicle;

[0038] The repulsive potential field generated by the position of the controlled vehicle;

[0039] The negative gradient of the repulsive potential field of the i-th environmental vehicle on the controlled vehicle;

[0040] The repulsive force generated by the position of the i-th environmental vehicle on the controlled vehicle represents The direction of fastest descent and the maximum rate of change at that point;

[0041] : The repulsive force generated by the i-th environmental vehicle on the controlled vehicle, representing The fastest falling direction and the maximum rate of change at that point;

[0042] The repulsive force generated by the speed of the i-th environmental vehicle on the controlled vehicle represents The direction of fastest descent and the maximum rate of change at that point.

[0043] Furthermore, in step S1, the road boundary repulsive potential field function of the road boundary repulsive potential field is constructed by: using the Euclidean distance between the controlled vehicle and the road boundary The maximum distance range of the repulsive force exerted by the road boundary on the controlled vehicle , determine the road boundary repulsive potential field function, that is:

[0044] ;

[0045] in, is the road boundary repulsion gain coefficient; The point representing the road boundary closest to the point where the controlled vehicle is located, i.e., the intersection of a perpendicular line passing through the controlled vehicle position to the road boundary; It is 1 / 2 of the road width, that is, the controlled vehicle is not considered to be affected by the road boundary repulsion only when it is located on the center line of the road; The repulsive potential field generated by the road boundary on the controlled vehicle.

[0046] Furthermore, the repulsive force function corresponding to the road boundary repulsive force potential field function is:

[0047] ;

[0048] in, Refers to the repulsive force generated by the road boundary on the controlled vehicle, which represents The direction of fastest descent and the maximum rate of change at that point.

[0049] Furthermore, in step S2, the method for generating the initial trajectory includes: using an artificial potential field method to obtain the force conditions of the controlled vehicle in a continuous time series, according to The kinematic relationship of is used to plan the vehicle path, determine the initial position of the vehicle, the target position and the position of obstacles in the environment, use the superposition of the gravitational field and the repulsive field to form a total potential field, and obtain the total force field of the controlled vehicle according to the total potential field gradient descent method and the predetermined speed model and update the vehicle position to generate a continuous trajectory; the generated trajectory is smoothed by the local weighted regression algorithm; among them, the kinematic relationship In , a is the kinematic acceleration, which here represents the acceleration of the controlled vehicle; is the first-order derivative of velocity in kinematics; is the second-order derivative of the kinematic position.

[0050] Furthermore, the specific method of using the local weighted regression algorithm to smooth the generated trajectory is as follows:

[0051] 1) Determine the local neighborhood: for a given point in the obtained trajectory data First, you need to determine the local neighborhood and select the bandwidth Determine the neighborhood distance The nearest data point, n is the total number of trajectories, and the set of data points in the neighborhood is ,in is the kth point in the neighborhood. (α is chosen to strike a balance between capturing local trends and preserving the original trajectory characteristics, effectively smoothing the trajectory without oversmoothing it).

[0052] 2) Calculate weights: For neighborhood Every point within , using the weight function to calculate its relative Weight , select Tricube weight function for calculation, , = , that is, the neighborhood Inside point The maximum distance, where Refers to the point where the trajectory position is to be smoothly optimized during trajectory smoothing processing; In the local area neighborhood All data points within To the destination The maximum value of the distance; In the local area neighborhood All data points within To the destination The maximum distance. (Reason for choosing the Tricube weight function: Compared with some other weight functions, the cubic weight function performs more moderately in suppressing the influence of distant points and highlighting the role of nearby points. It does not cause the weight to decay too drastically or too slowly, thus providing a more reasonable weight distribution for local weighted regression.)

[0053] 3) Local weighted regression: in the neighborhood Within, weighted linear regression is performed to estimate The corresponding smoothing value ;

[0054] Using linear models ,in is the error term, and the weighted sum of squared errors needs to be minimized:

[0055] ;

[0056] ;

[0057] .

[0058] Available and The smoothed value can be calculated , and then traverse each point of the original trajectory Repeat the above steps to get the smoothed trajectory .

[0059] Note: Subscript j indicates the index of the specific data point currently being processed or smoothed. Subscript i: used to determine When it is a local neighborhood of , it is used to represent the index of other points in the neighborhood; express local neighborhood The kth point in is the index of the point in the neighborhood, which is selected from the index i of the original data point, and the value of i ranges from 1 to n;

[0060] —The intercept term of the linear model, which represents the offset on the y-axis during linear fitting;

[0061] —The slope term of the linear model reflects the rate at which the independent variable x affects the dependent variable y;

[0062] —Neighborhood The kth point in Relative to the target point The weight is used for weighted regression to reflect the influence of the point on the regression estimate;

[0063] —Neighborhood The kth point in The corresponding original dependent variable data;

[0064] —The weighted sum of squared errors is used to measure the fitting effect of the linear model on the data in the neighborhood, and the optimal value is determined by minimizing it. ;

[0065] —The intercept term obtained by minimizing the weighted sum of squared errors estimated value of;

[0066] —The slope term obtained by minimizing the weighted sum of squared errors estimated value of;

[0067] —Target point The smoothed value obtained after local weighted regression is the output result after trajectory smoothing.

[0068] Furthermore, in step S3, the closed-loop control is implemented by: obtaining the state information of the controlled vehicle after trajectory generation and smoothing, including the position and speed of the controlled vehicle; feeding back the real-time position and speed information of the controlled vehicle to the controller, and at the same time enabling the controller to obtain the environmental vehicle information, combine the dynamic data of the environmental vehicles, periodically update the artificial potential field parameters and re-plan the trajectory, so as to adjust and optimize the subsequent vehicle trajectory according to the overall traffic conditions.

[0069] Compared with the prior art, the advantages of the present invention are:

[0070] By constructing a reasonable artificial potential field model and taking into account the impact of various factors on vehicle driving, the present invention can effectively improve traffic operation efficiency, reduce vehicle delays, reduce carbon emissions, and is environmentally friendly. It has a positive effect on alleviating traffic congestion in expressway weaving areas and improving road safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a flowchart of a trajectory planning method for a networked autonomous driving vehicle based on an artificial potential field method according to the present invention;

[0072] Figure 2 is a spatial distribution diagram of the gravitational potential field of the target point in an embodiment of the present invention;

[0073] Figure 3 : is a spatial distribution diagram of the road boundary repulsive potential field in an embodiment of the present invention;

[0074] Figure 4 is the spatial distribution of the combined field of the target point gravitational potential field, the dynamic environment vehicle repulsive potential field, and the road boundary repulsive potential field in the embodiment of the present invention;

[0075] Figure 5 This figure shows the trajectory generation process of the controlled vehicle under the action of the artificial potential field and the smoothing effect of the local weighted regression algorithm on the trajectory in an embodiment of the present invention. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be further described below.

[0077] This embodiment is based on the trajectory planning method of the networked autonomous driving vehicle using the artificial potential field method. Figure 1 As shown, the following steps are included:

[0078] S1: Based on the basic conditions of the expressway (including the construction of the expressway road boundary and the location of static obstacles on the expressway) and the traffic conditions (including the real-time location and speed of the surrounding vehicles), an artificial potential field is constructed, including:

[0079] 1) The target point gravitational potential field is generated based on the relative position difference and relative speed difference between the controlled vehicle and the target node;

[0080] In this embodiment, the method for constructing the target point gravitational potential field is: based on the relative position difference between the controlled vehicle and the target node and relative speed difference The target point attraction potential field can be regarded as the result of the merging area point moving along the target lane at the restricted speed. The target point attraction potential field function is defined as: ;

[0081] in, is the center of mass position of the controlled vehicle; is the target node position; is the controlled vehicle speed; The target node speed is the speed limit of the expressway; are the Euclidean distances of relative position and relative velocity respectively; 、 are the gain coefficients of target point position and velocity gravity respectively.

[0082] The gravitational function corresponding to the gravitational potential field function of the target point is the direction in which the potential field drops fastest, that is:

[0083] ;

[0084] in, Used to make the controlled vehicle track the position of the target node, The speed used to make the controlled vehicle track the target node, is the gravitational function corresponding to the gravitational potential field function of the target point, which represents the gravitational potential field function The fastest falling direction and the maximum rate of change at that point; Represents the negative gradient of the gravitational potential field.

[0085] like Figure 2 The spatial distribution of the gravitational potential field of the target point shown in the figure shows that through the lateral position, longitudinal position and gravitational potential field distribution of the expressway in the figure, it can be seen that the greater the distance between the controlled vehicle and the target point, the greater the potential energy value of the controlled vehicle; the smaller the distance, the smaller the potential energy value of the controlled object; the greater the speed difference, the greater the potential energy value of the controlled vehicle.

[0086] 2) The dynamic environment vehicle repulsive potential field generated based on the relative position and relative velocity of the controlled vehicle and the dynamic environment vehicle;

[0087] In this embodiment, the specific method for constructing the dynamic environment vehicle repulsive potential field in step S1 is: based on the relative position of the controlled vehicle and the i-th environment vehicle and relative speed , determine the dynamic environment vehicle repulsive potential field function as:

[0088] ;

[0089] ;

[0090] in, 、 is the position and speed of the i-th environment vehicle, 、 are the environmental vehicle position and velocity repulsion gain coefficients respectively, is the maximum range of the environmental vehicle;

[0091] The repulsive potential field generated by the i-th environmental vehicle on the controlled vehicle;

[0092] ):Right now , represents the vector difference between the position of the controlled vehicle and the position of the i-th vehicle;

[0093] :Right now , indicating that the speed of the i-th vehicle is a vector difference with the speed of the i-th vehicle.

[0094] In the dynamic environment vehicle repulsive potential field function:

[0095] when ;

[0096] when hour, ;

[0097] when hour, ;

[0098] when hour, ;

[0099] The repulsive force function corresponding to the repulsive force potential field function of the dynamic environment vehicle is:

[0100] ;

[0101] in, The repulsive potential field generated by the velocity of the i-th environmental vehicle on the controlled vehicle;

[0102] The repulsive potential field generated by the position of the i-th environmental vehicle on the controlled vehicle;

[0103] The negative gradient of the repulsive potential field of the i-th environmental vehicle on the controlled vehicle;

[0104] The repulsive force generated by the position of the i-th environmental vehicle on the controlled vehicle represents The direction of fastest descent and the maximum rate of change at that point;

[0105] represents the repulsive force generated by the i-th environmental vehicle on the controlled vehicle, and represents The fastest falling direction and the maximum rate of change at that point;

[0106] The repulsive force generated by the speed of the i-th environmental vehicle on the controlled vehicle represents The direction of fastest descent and the maximum rate of change at that point.

[0107] like Figure 4 As shown in the figure, it is the spatial distribution of the combined field of the target point gravitational potential field, the dynamic environment vehicle repulsive potential field, and the road boundary repulsive potential field. The potential field in each area increases as it approaches obstacles and road boundaries; the potential field decreases as it approaches the target point, and increases as it moves away from the target point. The controlled vehicle will choose the route direction where the potential field decreases the fastest.

[0108] when When , it means that the relative position vector and relative velocity vector of the controlled vehicle and the environment vehicle are less than 90°, that is, the relative velocity will increase the risk of collision;

[0109] when When , it means that the relative position vector and relative velocity vector of the controlled vehicle and the environment vehicle are greater than 90°, that is, the relative velocity will reduce the risk of collision.

[0110] 3) The road boundary repulsive potential field generated based on the Euclidean distance between the controlled vehicle and the road boundary;

[0111] In this embodiment, the method for constructing the road boundary repulsive potential field function in step S1 is: the Euclidean distance between the controlled vehicle and the road boundary is calculated. The maximum range of repulsion exerted by the road boundary on the controlled vehicle Determine the road boundary repulsive potential field function, namely:

[0112] ;

[0113] in, is the road boundary repulsion gain coefficient; The point representing the road boundary closest to the point where the controlled vehicle is located, i.e., the intersection of a perpendicular line passing through the controlled vehicle position to the road boundary; It is 1 / 2 of the road width, that is, the controlled vehicle is not considered to be affected by the road boundary repulsion only when it is located on the center line of the road; The repulsive potential field generated by the road boundary on the controlled vehicle.

[0114] The repulsive function corresponding to the road boundary repulsive potential field function is:

[0115] ;

[0116] in, Refers to the repulsive force generated by the road boundary on the controlled vehicle, representing The direction of fastest descent and the maximum rate of change at that point.

[0117] like Figure 3 The spatial distribution of the road boundary repulsive force potential field shown in Figure 1 shows that the greater the distance between the controlled vehicle and the road boundary, the smaller the repulsive force on the controlled vehicle; the smaller the distance, the greater the repulsive force on the controlled vehicle. When the distance between the controlled vehicle and the road boundary exceeds the effective distance, the road boundary potential field is zero.

[0118] S2: Based on the constructed artificial potential field, the initial vehicle trajectory is generated by calculating the resultant force (force conditions) of the controlled vehicle in the potential field, and the initial trajectory is smoothed using the Loess algorithm to reduce trajectory fluctuations;

[0119] In this embodiment, the method for generating the initial trajectory includes: using the artificial potential field method to obtain the force conditions of the controlled vehicle in a continuous time series, and according to (a is the kinematic acceleration, which here represents the acceleration of the controlled vehicle; is the first-order derivative of velocity in kinematics; The kinematic relationship of the vehicle (where the second-order derivative of the kinematic position is used) is used to plan the vehicle path, determine the vehicle's initial position, target position, and the positions of obstacles in the environment, and use the superposition of the gravitational field and the repulsive field to form a total potential field. The total force field acting on the controlled vehicle is obtained according to the total potential field gradient descent method and the predetermined speed model, and the vehicle position is updated to generate a continuous trajectory. The generated trajectory is then smoothed using a local weighted regression algorithm to reduce trajectory fluctuations and improve driving smoothness and ride comfort. Figure 5 Schematic diagram of the trajectory generation process of the controlled vehicle under the action of the artificial potential field and the smoothing effect of the local weighted regression algorithm on the trajectory in this embodiment.

[0120] S3: The controlled vehicle state after smoothing This information is fed back to the controller, which, combined with real-time data from surrounding vehicles, forms a closed-loop control system to dynamically adjust the vehicle's trajectory. The artificial potential field parameters in this closed-loop control are updated at regular intervals. Due to the complex environment of intersecting expressways in urban areas, the update frequency is 50Hz to ensure timely response to various dynamic changes.

[0121] In step S3, the closed-loop control is specifically implemented as follows: obtaining the state information of the controlled vehicle after trajectory generation and smoothing, including the position and speed of the controlled vehicle; feeding back the real-time position and speed information of the controlled vehicle to the controller, and at the same time enabling the controller to obtain information about the surrounding vehicles, combine the dynamic data of the surrounding vehicles, periodically update the artificial potential field parameters and replan the trajectory, thereby adjusting and optimizing the subsequent vehicle trajectory according to the overall traffic conditions.

[0122] The simulated traffic control experiment is carried out by the method of this embodiment, as follows:

[0123] 1) Experimental Scenario: A SUMO simulated road network consisting of six nodes and five edges was constructed. The main road was set to a total length of 1200 meters, and the distance between two ramps was 1000 meters. The maximum speed limit for the main lane was 80 km / h (equivalent to 22.2 m / s), while the speed limit for the ramps was fixed at 40 km / h (equivalent to 11.11 m / s).

[0124] 2) Experimental Parameters: The traffic flow was set to 2400 veh / h for straight-ahead traffic on the expressway, 1200 veh / h for merging onto ramps, and 800 veh / h for exiting the main road. The simulation captured data from 600 to 3600 s, excluding the first 600 s of warm-up time. The SUMO simulation configuration file was configured to output a tripinfo file to obtain vehicle delay data.

[0125] 3) Set up the control group and experimental group:

[0126] Control group: SUMO's built-in lane-changing model LC2013 and the car-following model Krauss's control method were used to simulate natural traffic flow conditions, henceforth referred to as "no control";

[0127] Experimental team: Control the simulation program through the Traci interface, including the construction of gravity and repulsion, and ensure that it can correctly interact with SUMO and achieve the expected functions;

[0128] The other kinematic parameters of the two groups are set to the same, including the maximum speed of 25m / s, the maximum acceleration of 2.6m / s2, the maximum deceleration of 4.5m / s2, and the body length is set to 5m.

[0129] 4) Simulation Experiment Results: The traffic flow path going straight on the main line is recorded as route 1, the traffic flow path entering the main line from the ramp is recorded as route 2, and the traffic flow path exiting the main line is recorded as route 3. The delay time data characteristics of each route are shown in Table 1 below:

[0130] Table 1:

[0131]

[0132] The above experiments show that based on the above weaving area control method, the delay of vehicles going straight on the main road is reduced by 31.3%, the delay of vehicles leaving the main road is reduced by 50.1%, and the delay of vehicles merging onto the ramp is reduced by 39.9%. This result shows that the present invention has a significant effect in improving the operating efficiency of the weaving area and reducing delays.

[0133] The above description is merely a preferred embodiment of the present invention and does not limit the present invention in any way. Any person skilled in the art who, without departing from the scope of the present invention, makes any equivalent substitution, modification, or other changes to the technical solution and technical content disclosed in the present invention shall be deemed to be within the scope of the present invention and still fall within the scope of protection of the present invention.

Claims

1. A trajectory planning method for a connected autonomous driving vehicle based on an artificial potential field method, characterized in that: The steps include: S1: Construct an artificial potential field based on the basic conditions and traffic conditions of the expressway, including: The target point gravitational potential field is generated by the position and velocity of the target point. The method for constructing the target point gravitational potential field is as follows: based on the relative position difference between the controlled vehicle and the target node and relative speed difference , define the gravitational potential field function of the target point as: ;in, is the center of mass position of the controlled vehicle; is the target node position; is the controlled vehicle speed; The target node speed is the speed limit of the expressway; are the Euclidean distances of relative position and relative velocity, 、 are the gain coefficients of the target point position and velocity gravity respectively; The dynamic environment vehicle repulsion potential field is generated by the dynamic environment vehicle position and speed. The dynamic environment vehicle repulsion potential field is constructed by: based on the relative position of the controlled vehicle and the i-th environment vehicle and relative speed , determine the dynamic environment vehicle repulsive potential field function as: ； ； in, 、 is the position and speed of the i-th environment vehicle, 、 are the environmental vehicle position and velocity repulsion gain coefficients respectively, is the maximum range of the environmental vehicle; : the repulsive potential field generated by the i-th environmental vehicle on the controlled vehicle; ):Right now , represents the vector difference between the position of the controlled vehicle and the position of the i-th vehicle; :Right now , indicating that the speed of the i-th vehicle is a vector difference with the speed of the i-th vehicle; The road boundary repulsion potential field generated by the road boundary, the road boundary repulsion potential field function of the road boundary repulsion potential field is constructed by comparing the Euclidean distance between the controlled vehicle and the road boundary The maximum range of repulsion exerted on the controlled vehicle by the road boundary , construct the road boundary repulsive potential field function: ； in, is the road boundary repulsion gain coefficient; The repulsive potential field generated by the road boundary on the controlled vehicle; S2: Based on the artificial potential field, the initial trajectory of the vehicle is generated by calculating the resultant force of the controlled vehicle in the potential field, and the initial trajectory is smoothed; the method for generating the initial trajectory includes: using the artificial potential field method to obtain the force situation of the controlled vehicle in a continuous time series, and according to The kinematic relationship of is used to plan the vehicle path, determine the vehicle's initial position, target position, and obstacle positions in the environment, and use the gravitational field and repulsive field to superimpose to form a total potential field. The total force field on the controlled vehicle is obtained according to the total potential field gradient descent method and the predetermined velocity model, and the vehicle position is updated to generate a continuous trajectory. The generated trajectory is smoothed using a local weighted regression algorithm. Where a is the kinematic acceleration, which here represents the acceleration of the controlled vehicle. is the first-order derivative of velocity in kinematics; is the second-order derivative of position in kinematics; S3: Feedback the smoothed state information of the controlled vehicle to the controller, combine it with the real-time data of the surrounding vehicles to form a closed-loop control, and dynamically adjust the vehicle trajectory. The closed-loop control is implemented by: obtaining the state information of the controlled vehicle after trajectory generation and smoothing, including the position and speed of the controlled vehicle; feeding back the real-time position and speed information of the controlled vehicle to the controller, and at the same time allowing the controller to obtain the surrounding vehicle information, combine it with the dynamic data of the surrounding vehicles, periodically update the artificial potential field parameters and re-plan the trajectory, so as to adjust and optimize the subsequent vehicle trajectory according to the overall traffic conditions.

2. The trajectory planning method for a networked autonomous driving vehicle based on an artificial potential field method according to claim 1, characterized in that: The gravitational function corresponding to the gravitational potential field function of the target point is the direction in which the potential field drops fastest, that is, ;in, Used to make the controlled vehicle track the position of the target node, The speed used to make the controlled vehicle track the target node, is the gravitational function corresponding to the gravitational potential field function of the target point, which represents the gravitational potential field function The fastest falling direction and the maximum rate of change at that point; is the negative gradient of the gravitational potential field.

3. The trajectory planning method for a networked autonomous driving vehicle based on an artificial potential field method according to claim 1, characterized in that: In the dynamic environment vehicle repulsive potential field function: when ; when hour, ; when hour, ; when hour, ; The repulsive force function corresponding to the dynamic environment vehicle repulsive force potential field function is: ； in, : The repulsive potential field generated by the velocity of the i-th environmental vehicle on the controlled vehicle; The repulsive potential field generated by the position of the i-th environmental vehicle on the controlled vehicle; The negative gradient of the repulsive potential field of the i-th environmental vehicle on the controlled vehicle; The repulsive force generated by the position of the i-th environmental vehicle on the controlled vehicle represents The direction of fastest descent and the maximum rate of change at that point; : The repulsive force generated by the i-th environmental vehicle on the controlled vehicle, representing The fastest falling direction and the maximum rate of change at that point; The repulsive force generated by the speed of the i-th environmental vehicle on the controlled vehicle represents The direction of fastest descent and the maximum rate of change at that point.

4. The trajectory planning method for a networked autonomous driving vehicle based on an artificial potential field method according to claim 1, characterized in that: The repulsive force function corresponding to the road boundary repulsive force potential field function is: ; in, Refers to the repulsive force generated by the road boundary on the controlled vehicle, which represents The direction of fastest descent and the maximum rate of change at that point.

5. The trajectory planning method for a networked autonomous driving vehicle based on an artificial potential field method according to claim 1, characterized in that: The specific implementation method of using the local weighted regression algorithm to smooth the generated trajectory includes: 1) Determine the local area and set the data points in the neighborhood as ,in, is the kth point in the neighborhood; 2) Calculate weights: For neighborhood Every point within , use the Tricube weight function to calculate its relative Weight , = , that is, the neighborhood Inside point The maximum distance; where Refers to the point where the trajectory position is to be smoothly optimized during trajectory smoothing processing; In the local area neighborhood All data points within To the destination The maximum value of the distance; In the local area neighborhood All data points within To the destination The maximum value of the distance; 3) Local weighted regression: in the neighborhood Within, weighted linear regression is performed to estimate Corresponding smoothing value .

Images