Methods and related equipment for predicting the lane change trajectory of autonomous vehicles on highways

By combining the Stackelberg game model and imitation learning algorithm with the MPC algorithm, interaction pairs between autonomous vehicles and surrounding vehicles are generated, and the optimal lane-changing trajectory is calculated. This solves the problem of large prediction errors in existing technologies and improves the driving safety and transportation efficiency of autonomous vehicles.

CN116729431BActive Publication Date: 2026-11-17SHENZHEN HANGSHENG ELECTRONICS
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202310642381.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-05-31
Publication Date
2026-11-17
Estimated Expiration
2043-05-31

AI Technical Summary

Technical Problem

Existing methods for predicting the trajectory of autonomous vehicles changing lanes on highways cannot accurately handle the random states of vehicle motion, resulting in large prediction errors and failing to effectively optimize vehicle driving safety and transportation efficiency.

Method used

The Stackelberg game model is used to generate interaction pairs between the target vehicle and surrounding vehicles. By combining imitation learning and the MPC algorithm, the optimal lane-changing trajectory that satisfies the preset safety strategy is calculated, taking into account the dynamic interaction of vehicles and safety constraints.

Benefits of technology

It improves the accuracy and safety of vehicle trajectory prediction, enhances the driving safety and transportation efficiency of autonomous vehicles, and solves the problems of large prediction errors and low efficiency in existing technologies.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116729431B_ABST
    Figure CN116729431B_ABST
Patent Text Reader

Abstract

This invention discloses a method and related equipment for predicting the lane-changing trajectory of an autonomous vehicle on a highway. The method includes: collecting state information data of the autonomous target vehicle and its surrounding environment; generating interaction pairs between the target vehicle and surrounding vehicles based on the state information data using a Stackelberg game model; and calculating the optimal lane-changing trajectory that satisfies a preset safety strategy based on the interaction pairs between the target vehicle and surrounding vehicles. This invention solves the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy; it considers the interaction between the autonomous vehicle and surrounding vehicles, handling the uncertainty of the interaction, and addresses the problem of most current decision-making models lacking interaction between the autonomous vehicle and surrounding vehicles. Furthermore, it solves the relevant constraints of the autonomous vehicle's safety limitations using optimization problems, thereby improving the correctness, success rate, and safety of the decision-making process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a method and related equipment for predicting the lane-changing trajectory of an autonomous vehicle on a highway. Background Technology

[0002] In complex hybrid driving environments characterized by the interaction and interpenetration of autonomous and human-driven vehicles, vehicles exhibit independence, constraints, and transmissibility. For a considerable period to come, a mixed traffic environment will exist where traditional human-driven vehicles and autonomous vehicles coexist. Research on the behavior of autonomous vehicles must fully consider urban environments and traffic scenarios with higher requirements for real-time reliable information and controllable driving assistance.

[0003] Existing key theories and methods for lane changing in autonomous vehicles suffer from problems such as homogeneity of vehicles, idealization of time, space and task constraints, making it difficult to achieve optimal solutions for vehicle driving safety and transportation efficiency.

[0004] Furthermore, current methods for predicting lane-changing trajectories in highway autonomous driving primarily employ methods based on fixed kinematic models. This involves discretely observing the target vehicle's motion state across multiple consecutive sampling periods within an observation cycle. For example, Chinese patent CN202211593201 is based on this method. However, this method suffers from the following technical drawbacks:

[0005] 1. It is impossible to model the random state of the target vehicle, meaning that the vehicle motion observation relies entirely on the dynamic model.

[0006] 2. The first trajectory representation sequence is input into the vehicle trajectory prediction model to optimize the data of each trajectory point in the second trajectory representation sequence. There is no feedback correction link. If there is an error in the first trajectory prediction or observation result, the vehicle control will have a large deviation.

[0007] Therefore, existing methods for predicting lane change trajectories in autonomous driving on highways have low accuracy in predicting vehicle trajectories, making it difficult to achieve optimal solutions for vehicle driving safety and transportation efficiency. Summary of the Invention

[0008] The main objective of this invention is to provide a method, system, terminal device, and storage medium for predicting the trajectory of autonomous vehicles changing lanes on highways, aiming to improve the accuracy of vehicle trajectory prediction and enhance vehicle driving safety and transportation efficiency.

[0009] To achieve the above objectives, embodiments of the present invention provide a method for predicting the trajectory of an autonomous vehicle changing lanes on a highway, the method comprising the following steps:

[0010] Collect status information data of the target autonomous vehicle and its surrounding environment;

[0011] Based on the state information data, interaction pairs between the target vehicle and surrounding vehicles are generated using the Stackelberg game model.

[0012] Based on the interaction pairs between the target vehicle and surrounding vehicles, the optimal lane-changing trajectory that satisfies the preset safety strategy is calculated.

[0013] Optionally, the step of collecting state information data of the autonomous driving target vehicle and its surrounding environment includes:

[0014] The system collects real-time status information data of the target vehicle and its surrounding environment through sensors, and obtains the target vehicle's speed, acceleration, heading angle, and center of gravity sideslip angle, as well as the speed, acceleration, and heading angle of surrounding vehicles.

[0015] Optionally, the step of generating the interaction pairs between the target vehicle and surrounding vehicles based on the state information data and the Stackelberg game model includes:

[0016] Based on the state information data, the target vehicle is configured with various accelerations, and multiple trajectories are generated using a fifth-order polynomial.

[0017] By combining the generated multiple trajectories, an interaction model between the target vehicle and surrounding vehicles is established based on the Stackelberg game model. The target vehicle is identified as the leader vehicle, and the other vehicles are the followers. Interaction pairs between the target vehicle and surrounding vehicles are generated.

[0018] Optionally, the step of calculating the optimal lane-changing trajectory that satisfies the preset safety strategy based on the interaction pairs between the target vehicle and surrounding vehicles includes:

[0019] Design a reward function for the interaction pair;

[0020] The observation vectors of the interaction pairs are generated based on the reward function;

[0021] Imitation learning is used to calculate the number and probability of lane change trajectories generated by the interaction pairs;

[0022] Based on the number and probability of generated lane change trajectories from the interaction, safety constraints are designed, and the MPC algorithm is used to find the optimal lane change trajectory that satisfies the optimal safety strategy.

[0023] Optionally, the method further includes:

[0024] Determine whether the optimal trajectory satisfies vehicle dynamics constraints;

[0025] The decision to change lanes or abandon the plan will be made based on the assessment results.

[0026] Optionally, the method further includes:

[0027] The state of the target vehicle is updated in a continuous space to obtain data for the next cycle.

[0028] Optionally, the reward function includes one or more of the following parameters: safety parameters, target state of the following vehicle, and comfort parameters. The safety constraints include one or more of the following constraints: design speed, acceleration, heading angle, and center of gravity sideslip angle.

[0029] This invention also proposes a system for predicting the lane-changing trajectory of an autonomous vehicle on a highway, the system comprising:

[0030] The data acquisition module is used to collect status information data of the target autonomous vehicle and its surrounding environment.

[0031] The generation module is used to generate interaction pairs between the target vehicle and surrounding vehicles based on the state information data and the Stackelberg game model.

[0032] Based on the interaction pairs between the target vehicle and surrounding vehicles, the optimal lane-changing trajectory that satisfies the preset safety strategy is calculated.

[0033] The present invention also proposes a terminal device, the terminal device including a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, it implements the method for predicting the highway lane change trajectory of an autonomous vehicle as described above.

[0034] The present invention also proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method for predicting the highway lane-changing trajectory of an autonomous vehicle as described above.

[0035] This invention proposes a method, system, terminal device, and storage medium for predicting the lane-changing trajectory of autonomous vehicles on highways. The method involves collecting state information data of the target autonomous vehicle and its surrounding environment; generating interaction pairs between the target vehicle and surrounding vehicles based on a Stackelberg game model; and calculating the optimal lane-changing trajectory that satisfies a preset safety strategy based on these interaction pairs. This invention utilizes a game model for vehicle prediction, addressing the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy. Furthermore, by considering the interaction between the vehicle and surrounding vehicles, it can handle the uncertainty of the interaction, solving the problem of most current decision-making models lacking interaction between the vehicle and surrounding vehicles. It also uses optimization problems to solve the relevant constraints of the vehicle's safety limitations, improving the correctness, success rate, and safety of the decision. Moreover, by using continuous state space solving, it addresses the problem of high computational complexity in discrete state space calculations in some current decision-making models, improving solution efficiency. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the functional modules of the terminal equipment belonging to the system for predicting the highway lane change trajectory of autonomous vehicles according to the present invention.

[0037] Figure 2 This is a flowchart illustrating an exemplary embodiment of the method for predicting the highway lane-changing trajectory of an autonomous vehicle according to the present invention.

[0038] Figure 3 This is a flowchart illustrating another exemplary embodiment of the method for predicting the highway lane-changing trajectory of an autonomous vehicle according to the present invention.

[0039] Figure 4 This is a flowchart illustrating yet another exemplary embodiment of the method for predicting the highway lane-changing trajectory of an autonomous vehicle according to the present invention.

[0040] Figure 5 This is a schematic diagram of the overall process of an embodiment of the method for predicting the lane change trajectory of an autonomous vehicle on a highway according to the present invention.

[0041] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0042] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0043] The main solution of this invention is as follows: First, collect state information data of the target autonomous vehicle and its surrounding environment. Second, based on the state information data, generate interaction pairs between the target vehicle and surrounding vehicles using a Stackelberg game model. Third, calculate the optimal lane-changing trajectory that satisfies a preset safety strategy based on the interaction pairs between the target vehicle and surrounding vehicles. This invention utilizes a game model for vehicle prediction, addressing the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy. Furthermore, by considering the interaction between the vehicle and surrounding vehicles, it can handle the uncertainty of the interaction, solving the problem of no interaction between the vehicle and surrounding vehicles in most current decision-making models. It solves the relevant constraints of the vehicle's safety limitations using optimization problems, improving the correctness, success rate, and safety of the decision. Finally, it uses continuous state space solving to address the high computational complexity of discrete state space calculations in some current decision-making models, improving solution efficiency.

[0044] Technical terms involved in the embodiments of this invention:

[0045] The Stackelberg game model, also known as the Stackelberg game, is a two-stage, fully informed dynamic game where the time frame is sequential. The main idea is that both players choose their own strategies based on the possible strategies of their opponents to maximize their own profit under those strategies, ultimately reaching a Nash equilibrium. In this game model, the player who makes the first decision is called the leader. Following the leader, the remaining players make decisions based on the leader's decisions and are called followers. The leader then adjusts its own decisions based on the followers' decisions, and this process continues until a Nash equilibrium is reached.

[0046] MPC algorithm: MPC stands for Model Predictive Control.

[0047] MPC, or Model Predictive Control, offers significant advantages in vehicle motion tracking trajectory control. The MPC algorithm is a multivariable control strategy involving: the dynamic model of the inner-loop process, historical values ​​of the control variable, and an optimal value equation over the prediction interval. The most significant characteristic of the MPC algorithm is that, compared to LQR control, MPC can consider various constraints on the spatial state variables, while LQR, PID, and other controls can only consider constraints on the input and output variables.

[0048] The working mechanism of the MPC algorithm is as follows: At each sampling time, based on the obtained current measurement information, a finite-time open-loop optimization problem is solved online, and the first element of the resulting control sequence is applied to the controlled object. At the next sampling time, the above process is repeated: the new measurement value is used as the initial condition for predicting the future dynamics of the system, the optimization problem is refreshed, and the solution is re-solved.

[0049] The MPC algorithm consists of three steps: predicting the future dynamics of the system, (numerically) solving the open-loop optimization problem, and applying the first element (or the first part) of the optimization solution to the system. These three steps are repeated at each sampling time, and regardless of the model used, the measurement value obtained at each sampling time serves as the initial condition for predicting the future dynamics of the system at the current time.

[0050] The MPC algorithm includes the following three basic principles (three basic characteristics):

[0051] Predictive models: Predictive models are the foundation of model predictive control. Their main function is to predict the future output of the system based on the static information of the object and future inputs. There are no strict limitations on the form of the predictive model; traditional models such as state equations and transfer functions can all be used as predictive models. For linear stable systems, nonparametric models such as step responses and impulse responses can also be used directly as predictive models.

[0052] Rolling optimization: Model predictive control determines the control action by finding the optimal performance index, but the optimization is not performed offline once, but repeatedly online. This is the meaning of rolling optimization, and it is also the fundamental difference between model predictive control and traditional optimal control.

[0053] Feedback correction: In order to prevent the control from deviating from the ideal state due to model mismatch or environmental interference, at a new sampling time, the actual output of the object is first detected, and this real-time information is used to correct the model-based prediction results before new optimization is performed.

[0054] This invention takes into account that in existing related solutions, in the complex environment of mixed driving (human / unmanned driving) where autonomous vehicles and human-driven vehicles interact and permeate each other, vehicles possess independence, constraints, and transmissibility. For a considerable period in the future, a mixed traffic environment will exist where traditional human-driven vehicles and autonomous vehicles coexist. Research on autonomous vehicle behavior needs to fully consider urban environments and traffic scenarios with higher requirements for real-time reliable information and controllable driving assistance. Existing key theories and methods for lane changing in autonomous vehicles suffer from problems such as homogeneity of vehicles, idealization of time, space, and task constraints, making it difficult to achieve optimal solutions for vehicle driving safety and transportation efficiency.

[0055] Therefore, the solutions proposed in this invention mainly address the following problems:

[0056] 1. Utilize game theory models for vehicle prediction to address the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy.

[0057] 2. It takes into account the interaction between the vehicle and surrounding vehicles, which can handle the uncertainty of the interaction and solve the problem that most current decision-making models do not have interaction between the vehicle and surrounding vehicles. It uses optimization problems to solve the relevant constraints of the vehicle's safety restrictions, thereby improving the correctness, success rate and safety of the decision.

[0058] 3. Using continuous state space to solve the problem of high computational complexity in discrete state space for some current decision models, thus improving the solution efficiency.

[0059] Specifically, refer to Figure 1 , Figure 1 This is a schematic diagram of the functional modules of the system to which the autonomous vehicle highway lane change trajectory prediction device of the present invention belongs. This autonomous vehicle highway lane change trajectory prediction device can be a data processing device independent of the terminal device, and it can be carried on the terminal device in hardware or software form. The terminal device can be a smart mobile terminal such as a mobile phone or tablet, or a network device such as a server.

[0060] In this embodiment, the terminal device of the autonomous vehicle highway lane change trajectory prediction device includes at least an output module 110, a processor 120, a memory 130, and a communication module 140.

[0061] The memory 130 stores the operating system and the highway lane-changing trajectory prediction program for autonomous vehicles; the output module 110 can be a display screen, speaker, etc. The communication module 140 can include a WIFI module, a mobile communication module, and a Bluetooth module, etc., and communicates with external devices or servers through the communication module 140.

[0062] In one embodiment, when the autonomous vehicle highway lane change trajectory prediction program in memory 130 is executed by the processor, it performs the following steps:

[0063] Collect status information data of the target autonomous vehicle and its surrounding environment;

[0064] Based on the state information data, interaction pairs between the target vehicle and surrounding vehicles are generated using the Stackelberg game model.

[0065] Based on the interaction pairs between the target vehicle and surrounding vehicles, the optimal lane-changing trajectory that satisfies the preset safety strategy is calculated.

[0066] Furthermore, when the autonomous vehicle highway lane-changing trajectory prediction program in memory 130 is executed by the processor, it also performs the following steps:

[0067] The system collects real-time status information data of the target vehicle and its surrounding environment through sensors, and obtains the target vehicle's speed, acceleration, heading angle, and center of gravity sideslip angle, as well as the speed, acceleration, and heading angle of surrounding vehicles.

[0068] Furthermore, when the autonomous vehicle highway lane-changing trajectory prediction program in memory 130 is executed by the processor, it also performs the following steps:

[0069] Based on the state information data, the target vehicle is configured with various accelerations, and multiple trajectories are generated using a fifth-order polynomial.

[0070] By combining the generated multiple trajectories, an interaction model between the target vehicle and surrounding vehicles is established based on the Stackelberg game model. The target vehicle is identified as the leader vehicle, and the other vehicles are the followers. Interaction pairs between the target vehicle and surrounding vehicles are generated.

[0071] Furthermore, when the autonomous vehicle highway lane-changing trajectory prediction program in memory 130 is executed by the processor, it also performs the following steps:

[0072] Design a reward function for the interaction pair;

[0073] The observation vectors of the interaction pairs are generated based on the reward function;

[0074] Imitation learning is used to calculate the number and probability of lane change trajectories generated by the interaction pairs;

[0075] Based on the number and probability of generated lane change trajectories from the interaction, safety constraints are designed, and the MPC algorithm is used to find the optimal lane change trajectory that satisfies the optimal safety strategy.

[0076] Furthermore, when the autonomous vehicle highway lane-changing trajectory prediction program in memory 130 is executed by the processor, it also performs the following steps:

[0077] Determine whether the optimal trajectory satisfies vehicle dynamics constraints;

[0078] The decision to change lanes or abandon the plan will be made based on the assessment results.

[0079] Furthermore, when the autonomous vehicle highway lane-changing trajectory prediction program in memory 130 is executed by the processor, it also performs the following steps:

[0080] The state of the target vehicle is updated in a continuous space to obtain data for the next cycle.

[0081] This embodiment, through the above-described scheme, specifically collects state information data of the target autonomous vehicle and its surrounding environment; based on the state information data, it generates interaction pairs between the target vehicle and surrounding vehicles using a Stackelberg game model; and based on these interaction pairs, it calculates the optimal lane-changing trajectory that satisfies a preset safety strategy. This invention utilizes a game model for vehicle prediction, addressing the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy. Furthermore, by considering the interaction between the vehicle and surrounding vehicles, it can handle the uncertainty of the interaction, solving the problem of most current decision-making models lacking interaction between the vehicle and surrounding vehicles. It solves the relevant constraints of the vehicle's safety limitations using optimization problems, improving the correctness, success rate, and safety of the decision. Moreover, it uses continuous state space solving, addressing the problem of high computational complexity in discrete state space calculations in some current decision-making models, thus improving solution efficiency.

[0082] Based on, but not limited to, the above system architecture, embodiments of the method of the present invention are proposed.

[0083] Reference Figure 2 , Figure 2 This is a flowchart illustrating an exemplary embodiment of the highway lane-change trajectory prediction method for autonomous vehicles according to the present invention. The method for predicting the highway lane-change trajectory of autonomous vehicles includes the following steps:

[0084] Step S101: Collect status information data of the autonomous driving target vehicle and its surrounding environment;

[0085] The method in this embodiment can be applied to the driving scenario of autonomous vehicles on highways, mainly to predict the lane-changing trajectory of autonomous vehicles on highways.

[0086] The subject executing the method in this embodiment can be a vehicle lane change trajectory prediction system.

[0087] In order to predict the lane-changing trajectory of autonomous vehicles on highways, it is necessary to first collect state information data of the target autonomous vehicle and its surrounding environment.

[0088] Specifically, various sensors installed on the target vehicle can collect real-time status information data of the target vehicle and its surrounding environment to obtain the target vehicle's speed, acceleration, heading angle, and center of gravity sideslip angle, as well as the speed, acceleration, and heading angle of surrounding vehicles.

[0089] Step S102: Based on the state information data, generate the interaction pairs between the target vehicle and surrounding vehicles using the Stackelberg game model;

[0090] Specifically, this embodiment uses a game theory model for vehicle prediction, which solves the problem that current mainstream decision-making models cannot improve prediction accuracy by using fixed dynamics model methods.

[0091] This process involves configuring various acceleration settings for the target vehicle and generating multiple trajectories using a fifth-order polynomial. Furthermore, an interaction model between the target vehicle and surrounding vehicles is established based on a Stackelberg game model, identifying the target vehicle as the leader and the other vehicles as followers, thus generating interaction pairs.

[0092] The acceleration model for the target vehicle with various acceleration configurations is as follows: β = 0;

[0093] Where x is the longitudinal position of the vehicle, y is the lateral position of the vehicle, v is the velocity of the vehicle, a is the acceleration of the vehicle, ψ is the yaw angle of the vehicle, and β is the sideslip angle of the vehicle.

[0094] The fifth-order polynomial trajectory model is:

[0095] x(t) = a0 + a1t + a2t 2 +a3t 3 +a4t 4 +a5t 5 ;

[0096] y(t)=b0+b1t+b2t 2 +b3t 3 +b4t 4 +b5t 5 ;

[0097] Where t is the path planning lane change cycle time, which is 2s in this scheme, a0, a1, a2, a3, a4, a5 are the coefficients of the fifth-order polynomial of the vehicle's longitudinal trajectory, and b0, b1, b2, b3, b4, b5 are the coefficients of the fifth-order polynomial of the vehicle's lateral trajectory.

[0098] The Stackelberg game Leader-Follower model is as follows:

[0099] Reward function:

[0100]

[0101]

[0102] σ∈L={Leader,Follower}

[0103] In this system, the vehicle that has priority over other vehicles when changing lanes is the Leader, and the vehicle that yields to other vehicles is the Follower.

[0104] In the formula, γ l,t It is a collection of Leader's trajectories, γ f,t It is a collection of Follower trajectories, γ l Current traffic status The optimal trajectory of the lower leader, γ f Current traffic status The optimal trajectory for the Follower.

[0105] λ is the reward function adjustment factor, with a value range of 0.0-1.0.

[0106] Q l Current traffic status The minimum cost function of the lower leader, and γ l The connection is: to make a decision that maximizes reward under the current traffic conditions using the minimum cost (finding the minimum) (finding Q). l Among all minimum cost trajectories, argmax), Q f Current traffic status The minimum cost function for the Follower.

[0107] The above methods generate the relationship between interactive pairs.

[0108] Specifically, as one implementation method, the step of generating the interaction pairs between the target vehicle and surrounding vehicles based on the state information data and the Stackelberg game model may include:

[0109] First, based on the state information data, the target vehicle is configured with various accelerations, and multiple trajectories are generated using a fifth-order polynomial;

[0110] By combining the generated multiple trajectories, an interaction model between the target vehicle and surrounding vehicles is established based on the Stackelberg game model. The target vehicle is identified as the leader vehicle, and the other vehicles are the follower vehicles. Interaction pairs between the target vehicle and surrounding vehicles are generated.

[0111] By using the Stackelberg game model to predict the trajectory of the target vehicle, the cooperative intentions of different interactions are taken into account, which can handle the uncertainty in the interaction.

[0112] Step S103: Based on the interaction pairs between the target vehicle and surrounding vehicles, calculate the optimal lane-changing trajectory that satisfies the preset safety strategy.

[0113] Specifically, to obtain the optimal lane-changing trajectory that satisfies the preset safety strategy, this embodiment designs a reward function for the interaction pair. This reward function may include: safety parameters, follow vehicle target state, comfort parameters, etc., and the interaction pair is vectorized. Imitation learning is used to generate the optimized number and probability of lane-changing trajectories by minimizing the reward function on the dataset.

[0114] The imitation learning formula can be expressed as:

[0115] ;in:

[0116] θ represents the weight parameters (labeled safety parameters, target state parameters, comfort parameters, etc.) in the imitation learning network, and γ represents the weight parameters. θ It is the policy value optimized based on the weight parameters in the imitation learning network. It is a dataset collection used in imitation learning. It is a subset of the dataset that conforms to the algorithm strategy in imitation learning. This represents the loss function in imitation learning.

[0117] Then, constraints on speed, acceleration, heading angle, and center of gravity sideslip angle can be designed, and the MPC algorithm can be used to calculate the lane change trajectory that satisfies the optimal safety strategy.

[0118] The MPC algorithm formula is as follows:

[0119] in:

[0120]

[0121]

[0122]

[0123] At each sampling time t, the vehicle calculates an optimal trajectory. Maximize the reward within the solution period. This represents the traffic state at the discrete time point t+τ. and λ and τ represent the predictive control inputs for the autonomous vehicle and the interactive vehicle at time t+τ, respectively. λ is the reward function adjustment factor, ranging from 0.0 to 1.0. S represents the reward obtained by the vehicle at time t+τ. safe These represent safety constraints (such as collision constraints, road edge constraints, etc.).

[0124] The vehicle calculates the optimal trajectory. Then, the vehicle calculates the control quantity based on the optimal trajectory. in It is the vehicle's acceleration value. It is the steering wheel angle value of the vehicle, and the status is updated at the same time.

[0125] Specifically, as one implementation method, the step of calculating the optimal lane-changing trajectory that satisfies the preset safety strategy based on the interaction pairs between the target vehicle and surrounding vehicles may include:

[0126] First, design a reward function for the interaction pair;

[0127] The reward function can include: safety parameters, follow vehicle target state, comfort parameters, etc.

[0128] Then, observation vectors for the interaction pairs are generated based on the reward function;

[0129] Then, imitation learning is used to calculate the number and probability of lane change trajectories generated by the interaction pair;

[0130] Finally, based on the number and probability of the generated lane change trajectories from the interaction, safety constraints are designed, and the MPC algorithm is used to obtain the optimal lane change trajectory that satisfies the optimal safety strategy.

[0131] The safety constraints may include one or more of the following: design speed, acceleration, heading angle, and center of mass sideslip angle constraints.

[0132] This embodiment, through the above-described scheme, specifically collects state information data of the target autonomous vehicle and its surrounding environment; based on the state information data, it generates interaction pairs between the target vehicle and surrounding vehicles using a Stackelberg game model; and based on these interaction pairs, it calculates the optimal lane-changing trajectory that satisfies a preset safety strategy. This invention utilizes a game model for vehicle prediction, addressing the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy. Furthermore, by considering the interaction between the vehicle and surrounding vehicles, it can handle the uncertainty of the interaction, solving the problem of most current decision-making models lacking interaction between the vehicle and surrounding vehicles. It also uses optimization problems to solve the relevant constraints of the vehicle's safety limitations, improving the correctness, success rate, and safety of the decision-making process.

[0133] Reference Figure 3 , Figure 3 This is a flowchart illustrating another exemplary embodiment of the highway lane change trajectory prediction method for autonomous vehicles according to the present invention.

[0134] Based on the above Figure 2 In the embodiment shown, the method further includes:

[0135] Step S104: Determine whether the optimal trajectory satisfies vehicle dynamics constraints;

[0136] Step S105: Determine whether to change lanes or abandon the decision based on the judgment result.

[0137] Compared to the above Figure 2 The embodiment shown also includes a scheme for controlling vehicle execution based on the optimal lane change trajectory.

[0138] Specifically, after obtaining the optimal lane-changing trajectory that satisfies the optimal safety strategy, it is determined whether the constraints of speed, acceleration, heading angle, and center of gravity sideslip angle meet the vehicle dynamics requirements. Then, based on the determination results, a decision is made on whether the vehicle should change lanes, such as changing lanes or abandoning the lane-changing decision.

[0139] Therefore, by incorporating vehicle dynamics safety constraints into an optimization problem, safety requirements can be met through optimization solutions.

[0140] This embodiment, through the above-described scheme, specifically collects state information data of the target autonomous vehicle and its surrounding environment; based on the state information data, it generates interaction pairs between the target vehicle and surrounding vehicles using a Stackelberg game model; and based on these interaction pairs, it calculates the optimal lane-changing trajectory that satisfies a preset safety strategy. This invention utilizes a game model for vehicle prediction, addressing the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy. Furthermore, by considering the interaction between the vehicle and surrounding vehicles, it can handle the uncertainty of the interaction, solving the problem of most current decision-making models lacking interaction between the vehicle and surrounding vehicles. It also uses optimization problems to solve the relevant constraints of the vehicle's safety limitations, improving the correctness, success rate, and safety of the decision-making process.

[0141] Reference Figure 4 , Figure 4 This is a flowchart illustrating another exemplary embodiment of the highway lane-changing trajectory prediction method for autonomous vehicles according to the present invention.

[0142] Based on the above Figure 4 In the embodiment shown, the method further includes:

[0143] Step S106: Update the state of the target vehicle in the continuous space to obtain data for the next cycle.

[0144] Compared to the above Figure 3 The embodiment shown also includes a scheme for updating the state of the target vehicle in a continuous space to obtain the data for the next cycle, thereby solving the problems of high computational complexity and high latency caused by discrete state space by using continuous state space.

[0145] This embodiment, through the above-described scheme, specifically collects state information data of the target autonomous vehicle and its surrounding environment; based on the state information data, it generates interaction pairs between the target vehicle and surrounding vehicles using a Stackelberg game model; and based on these interaction pairs, it calculates the optimal lane-changing trajectory that satisfies a preset safety strategy. This invention utilizes a game model for vehicle prediction, addressing the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy. Furthermore, by considering the interaction between the vehicle and surrounding vehicles, it can handle the uncertainty of the interaction, solving the problem of most current decision-making models lacking interaction between the vehicle and surrounding vehicles. It solves the relevant constraints of the vehicle's safety limitations using optimization problems, improving the correctness, success rate, and safety of the decision. Moreover, it uses continuous state space solving, addressing the problem of high computational complexity in discrete state space calculations in some current decision-making models, thus improving solution efficiency.

[0146] The specific process for predicting the highway lane-changing trajectory of an autonomous vehicle in this embodiment can be found in [reference needed]. Figure 5 As shown, it includes:

[0147] Step 1: The autonomous vehicle collects real-time status information data of itself and the surrounding environment through sensors, and obtains the vehicle's speed, acceleration, heading angle, and center of gravity sideslip angle, as well as the speed, acceleration, and heading angle of the surrounding vehicles.

[0148] Step 2: By configuring the vehicle with various acceleration settings, multiple trajectories are generated using a fifth-order polynomial. An interaction model between the vehicle and surrounding vehicles is established based on a Tuckberg game model, identifying the vehicle as the leader and the other vehicles as followers, thus generating interaction pairs.

[0149] Step 3: Design a reward function for each interaction pair. The reward function includes safety parameters, follow vehicle target state, and comfort parameters. Represent the interaction pair using vectorization. Use imitation learning to generate optimized lane-change trajectories and probabilities by minimizing the reward function on the dataset.

[0150] Step 4: Design constraints for speed, acceleration, heading angle, and sideslip angle, and use MPC to calculate the lane change trajectory that satisfies the optimal safety strategy.

[0151] Step 5: Determine whether the constraints of speed, acceleration, heading angle, and sideslip angle meet the vehicle's dynamic requirements, and decide whether to change lanes or abandon the change.

[0152] Step 6: Update the state in the continuous space to obtain the data for the next cycle.

[0153] Compared with existing technologies, the solution in this embodiment uses the Stackelberg game model to predict the trajectory of the target vehicle, which takes into account the cooperative intentions of different interactions and can handle the uncertainty in the interaction; it places the vehicle dynamics safety constraints in the optimization problem, and can meet the safety requirements through optimization solution; it uses continuous state space to solve the problems of high computational complexity and time delay caused by discrete state space.

[0154] Furthermore, embodiments of the present invention also propose a system for predicting the lane-changing trajectory of an autonomous vehicle on a highway, the system comprising:

[0155] The data acquisition module is used to collect status information data of the target autonomous vehicle and its surrounding environment.

[0156] The generation module is used to generate interaction pairs between the target vehicle and surrounding vehicles based on the state information data and the Stackelberg game model.

[0157] Based on the interaction pairs between the target vehicle and surrounding vehicles, the optimal lane-changing trajectory that satisfies the preset safety strategy is calculated.

[0158] The principle and implementation process of predicting the lane change trajectory of autonomous vehicles on highways in this embodiment are described in the above embodiments and will not be repeated here.

[0159] Furthermore, this embodiment of the invention also proposes a terminal device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the method for predicting the highway lane-changing trajectory of an autonomous vehicle as described in the above embodiments.

[0160] Since the autonomous vehicle highway lane change trajectory prediction program adopts all the technical solutions of all the aforementioned embodiments when it is executed by the processor, it has at least all the beneficial effects brought about by all the technical solutions of all the aforementioned embodiments, which will not be repeated here.

[0161] Furthermore, this embodiment of the invention also proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method for predicting the highway lane-changing trajectory of an autonomous vehicle as described in the above embodiments.

[0162] Since the autonomous vehicle highway lane change trajectory prediction program adopts all the technical solutions of all the aforementioned embodiments when it is executed by the processor, it has at least all the beneficial effects brought about by all the technical solutions of all the aforementioned embodiments, which will not be repeated here.

[0163] Compared to existing technologies, the method, system, terminal device, and storage medium for predicting highway lane-changing trajectories of autonomous vehicles proposed in this invention collect state information data of the autonomous target vehicle and its surrounding environment; based on the state information data, it generates interaction pairs between the target vehicle and surrounding vehicles using a Stackelberg game model; and based on the interaction pairs, it calculates the optimal lane-changing trajectory that satisfies a preset safety strategy. This invention utilizes a game model for vehicle prediction, addressing the problem that current mainstream decision-making models using fixed dynamics models cannot improve prediction accuracy. Furthermore, by considering the interaction between the vehicle and surrounding vehicles, it can handle the uncertainty of the interaction, solving the problem of most current decision-making models lacking interaction between the vehicle and surrounding vehicles. It solves the relevant constraints of the vehicle's safety limitations using optimization problems, improving the correctness, success rate, and safety of the decision. Moreover, it uses continuous state space solving, addressing the problem of high computational complexity in discrete state space calculations in some current decision-making models, thus improving solution efficiency.

[0164] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0165] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0166] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, controlled terminal, or network device, etc.) to execute the methods of each embodiment of the present invention.

[0167] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for predicting the lane-changing trajectory of an autonomous vehicle on a highway, characterized in that, The method includes the following steps: Collect status information data of the target autonomous vehicle and its surrounding environment; Based on the state information data, an interaction pair between the target vehicle and surrounding vehicles is generated using a Stackelberg game model. Specifically, this includes: configuring the target vehicle with various accelerations based on the state information data, generating multiple motion trajectories using a fifth-order polynomial; combining the generated multiple motion trajectories, establishing an interaction model between the target vehicle and surrounding vehicles based on the Stackelberg game model, determining the target vehicle as the lead vehicle and the other vehicles as follower vehicles, and generating an interaction pair between the target vehicle and surrounding vehicles. Based on the interaction pairs between the target vehicle and surrounding vehicles, the optimal lane-changing trajectory that satisfies the preset safety strategy is calculated. Specifically, this includes: designing a reward function for the interaction pairs; generating observation vectors for the interaction pairs based on the reward function; using imitation learning to calculate the number and probability of lane-changing trajectories generated by the interaction pairs; designing safety constraints based on the number and probability of lane-changing trajectories generated by the interaction pairs; and using the MPC algorithm to obtain the optimal lane-changing trajectory that satisfies the optimal safety strategy. The MPC algorithm calculates an optimal trajectory at each sampling time t, maximizing the reward within the solution period.

2. The method according to claim 1, characterized in that, The steps for collecting status information data of the autonomous driving target vehicle and its surrounding environment include: The system collects real-time status information data of the target vehicle and its surrounding environment through sensors, and obtains the target vehicle's speed, acceleration, heading angle, and center of gravity sideslip angle, as well as the speed, acceleration, and heading angle of surrounding vehicles.

3. The method according to claim 1, characterized in that, The method further includes: Determine whether the optimal trajectory satisfies vehicle dynamics constraints; The decision to change lanes or abandon the plan will be made based on the assessment results.

4. The method according to claim 3, characterized in that, The method further includes: The state of the target vehicle is updated in a continuous space to obtain data for the next cycle.

5. The method according to claim 1, characterized in that, The reward function includes one or more of the following parameters: safety parameters, target state of the following vehicle, and comfort parameters. The safety constraints include one or more of the following constraints: design speed, acceleration, heading angle, and center of gravity sideslip angle.

6. A system for predicting the trajectory of an autonomous vehicle changing lanes on a highway, characterized in that, The system includes: The data acquisition module is used to collect status information data of the target autonomous vehicle and its surrounding environment. The generation module is used to generate interaction pairs between the target vehicle and surrounding vehicles based on the state information data and the Stackelberg game model. Specifically, it includes: configuring the target vehicle with various accelerations based on the state information data, generating multiple motion trajectories using a fifth-order polynomial; combining the generated multiple motion trajectories, establishing an interaction model between the target vehicle and surrounding vehicles based on the Stackelberg game model, determining that the target vehicle is the leader vehicle and the other vehicles are following vehicles, and generating interaction pairs between the target vehicle and surrounding vehicles. Based on the interaction pairs between the target vehicle and surrounding vehicles, the optimal lane-changing trajectory that satisfies the preset safety strategy is calculated. Specifically, this includes: designing a reward function for the interaction pairs; generating observation vectors for the interaction pairs based on the reward function; using imitation learning to calculate the number and probability of lane-changing trajectories generated by the interaction pairs; designing safety constraints based on the number and probability of lane-changing trajectories generated by the interaction pairs; and using the MPC algorithm to obtain the optimal lane-changing trajectory that satisfies the optimal safety strategy. The MPC algorithm calculates an optimal trajectory at each sampling time t, maximizing the reward within the solution period.

7. A terminal device, characterized in that, The terminal device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When executed by the processor, the computer program implements the method for predicting the highway lane change trajectory of an autonomous vehicle as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for predicting the highway lane-changing trajectory of an autonomous vehicle as described in any one of claims 1-5.

Citation Information

Patent Citations

  • Vehicle trajectory prediction method, trajectory prediction model training method and device

    CN115743101A

  • Automatic lane changing planning method and device, electronic equipment and storage medium

    CN112835362A

  • Freight vehicle interactive game lane change decision-making method based on lane change decision-making system

    CN114882705A

  • Automatic driving vehicle lane changing trajectory planning method based on multi-person game

    CN116185027A