A method and system for trajectory control of an adaptive pruning intelligent connected vehicle

Abstract

The present invention discloses an adaptive pruning intelligent networked vehicle trajectory control method and system, wherein the method comprises: step S10, using a Monte Carlo tree search algorithm to generate an optimized trajectory plan for a target intelligent networked vehicle; step S20, pruning the action space of the trajectory plan, reducing the decision space by eliminating unsafe or unfeasible actions; step S30, according to the trajectory plan of the target intelligent networked vehicle, proposing optimal speed and lane change suggestions based on the current traffic conditions to other intelligent networked vehicles around it, and step S40, combining a low-level motion controller based on model predictive control to ensure that the intelligent networked vehicle drives smoothly in the action space generated by the trajectory plan. When the method is used, it overcomes the problem that the prior art usually focuses on vehicle following or collision avoidance behavior, and the control strategies between intelligent networked vehicles are mostly considered separately, lacking a systematic solution.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic monitoring, and in particular to a method and system for controlling trajectory of an adaptively pruned intelligent connected vehicle. Background Art

[0002] As urban traffic becomes increasingly congested, how to effectively manage mixed traffic flows and improve traffic efficiency and safety has become an urgent problem that needs to be solved. Mixed traffic flows refer to scenarios where automated vehicles and traditional vehicles share the road, which will be the norm in the next few decades. Key challenges in this context include how to improve the energy efficiency of the overall network without sacrificing traffic fluidity and safety. Especially in the context of the gradual commissioning of intelligent connected vehicles (CAVs), how to achieve effective interaction between CAVs and HDVs, optimize traffic flow, reduce traffic congestion, and improve fuel efficiency has become a research focus.

[0003] In an autonomous driving connected environment, CAVs have the potential to optimize their own driving paths and speeds and reduce energy consumption while ensuring safety through advanced sensors, communication technologies, and control algorithms. However, the potential of CAVs is not limited to self-optimization, but also includes influencing the behavior of the entire traffic flow through interactions with other CAVs around them. This influence can be achieved by providing speed and lane change recommendations to promote smoother and more energy-efficient traffic flow.

[0004] Existing technologies usually focus on vehicle following or collision avoidance behaviors, and the control strategies between CAVs are mostly considered separately. There is a lack of a systematic solution to achieve effective coordination between the two, thereby improving the efficiency and safety of the entire transportation system.

[0005] Therefore, it is an urgent problem to be solved by the present invention to provide an adaptive pruning intelligent connected vehicle trajectory control method and system that can overcome the above technical problems during use. Summary of the invention

[0006] In view of the above technical problems, the purpose of the present invention is to overcome the problem that the prior art usually focuses on vehicle following or collision avoidance behavior, and the control strategies between CAVs are mostly considered separately, lacking a systematic solution, so as to provide an adaptive pruning intelligent connected vehicle trajectory control method and system that can achieve effective coordination between the two during use, thereby improving the efficiency and safety of the entire transportation system.

[0007] In order to achieve the above object, the present invention provides an adaptive pruning intelligent connected vehicle trajectory control method based on Monte Carlo search, characterized in that the method comprises:

[0008] Step S10, using a Monte Carlo tree search algorithm to generate an optimized trajectory plan for the target intelligent connected vehicle;

[0009] Step S20, pruning the action space of trajectory planning, reducing the decision space by eliminating unsafe or infeasible actions, and improving the operation efficiency of the algorithm;

[0010] Step S30, according to the trajectory planning of the target intelligent networked vehicle, proposes optimal speed and lane change suggestions based on the current traffic conditions to other intelligent networked vehicles around it, so as to reduce traffic congestion and improve the overall efficiency of traffic flow;

[0011] Step S40, combined with a low-level motion controller based on model predictive control, ensures that the intelligent connected vehicle drives smoothly in the motion space generated by trajectory planning, while optimizing fuel efficiency and improving driving comfort.

[0012] Preferably, the step S10 includes the following steps:

[0013] Step S11, using optimal control to constrain intelligent connected vehicle control according to the following formula:

[0014]

[0015] It is subject to the dynamic constraints:

[0016] s x (k+1)=s x (k)+T s v(k)cos(θ(k))

[0017] s y (k+1)=s y (k)+T s v(k)sin(θ(k))

[0018] v(k+1)=v(k)+T s a(k)

[0019] θ(k+1)=θ(k)+T s ω(k)

[0020]

[0021] v min ≤v(τ)≤v max ,a min ≤a(τ)≤a max

[0022] Where k is a certain moment, τ is a certain moment in the control time domain, and s x (k) is the longitudinal position of the vehicle, s y(k) is the lateral position of the vehicle, v(k) is the vehicle speed, θ is the heading angle, a(k) is the vehicle acceleration, ω(k) is the vehicle steering acceleration, T s is the sampling time, is the fuel consumption rate, J is the other consumption of the trip, x(τ) is the trip distance, u(τ) is the formation time, S(τ) is the feasible area of ​​the road, S j (τ) is the set of positions occupied by road entities at time τ, is the set of entities around the vehicle, S safe (τ) is the safe driving area of ​​the vehicle at time τ.

[0023] Step S12, setting the reward function based on the Monte Carlo tree search algorithm:

[0024]

[0025] in:

[0026]

[0027]

[0028] Where: Δs is the longitudinal driving distance, is the fuel consumption of Δs, w1 is the fuel consumption weight, r c For collision reward, r v (τ) is close to the expected speed v des The reward is w2, and w3 is the acceleration / deceleration penalty weight.

[0029] Preferably, the fuel consumption rate in step S11 is The road test data is fitted with a cubic polynomial using the following formula:

[0030]

[0031] Preferably, the step S20 includes the following steps:

[0032] Step S21, dividing the vehicle behavior space into 14 combinations according to acceleration and steering wheel angle according to the following formula;

[0033]

[0034] Where: The first row of the matrix represents the acceleration (in m / s 2 ), the second row of the matrix represents the yaw angular velocity and is calculated in step S11;

[0035] Step S21, eliminating the unreasonable acceleration speed state by the following formula:

[0036]

[0037] Where: is the distance to the nearest object (vehicle / obstacle) in front of the intelligent connected vehicle in the same lane. is the object speed, w v3 ∈[0.8,1] is the linearly changing adaptive weight, and η2 is the time tolerance.

[0038] Preferably, the step S30 includes the following steps:

[0039] Step S31, using the following formula to plan the optimal speed for other surrounding intelligent connected vehicles;

[0040]

[0041] Where: q∈{lt,sm,r} represents left lane change, lane keeping and right lane change respectively. is the distance to the nearest road entity, 500 meters ahead at q th The average speed of all vehicles in the lane, d min is the minimum safety distance, T n is the headway parameter.

[0042] Preferably, the framework of the model predictive control based on which the low-level motion controller in step S40 is solved is solved by an optimized interior point method; and the motion controller based on the Monte Carlo tree search algorithm of the intelligent connected vehicle is implemented by the following formula:

[0043]

[0044] Where: w4 is the weighted acceleration for improving comfort, w5 is the penalty weight for trajectory deviation, s MCTS (τ) is the position of the intelligent connected vehicle obtained by the MCTS trajectory planner.

[0045] The present invention also provides an adaptive pruning intelligent connected vehicle trajectory control system based on Monte Carlo search, the system comprising:

[0046] Trajectory planning module, which is used to generate optimized trajectory plans for intelligent connected vehicles using the Monte Carlo tree search algorithm;

[0047] The trajectory optimization module is used to prune the action space of trajectory planning, reduce the decision space by eliminating unsafe or infeasible actions, and improve the operation efficiency of the algorithm;

[0048] The surrounding intelligent connected vehicle change suggestion module is used to propose optimal speed and lane change suggestions based on the current traffic conditions to other surrounding intelligent connected vehicles according to the trajectory planning of the target intelligent connected vehicle, so as to reduce traffic congestion and improve the overall efficiency of traffic flow;

[0049] The execution module is used to combine a low-level motion controller based on model predictive control to ensure that the intelligent connected vehicle drives smoothly in the action space generated by trajectory planning, while optimizing fuel efficiency and improving driving comfort.

[0050] According to the above technical solution, the adaptive pruning intelligent connected vehicle trajectory control method and system based on Monte Carlo search provided by the present invention have the following beneficial effects when used:

[0051] 1. This invention proposes a new control strategy for autonomous driving networked environment by combining Monte Carlo tree search algorithm (MCTS) algorithm and model predictive control (MPC) low-level motion controller, which can not only optimize the driving trajectory of intelligent networked vehicles, but also provide suggestions for other surrounding intelligent networked vehicles to achieve effective collaboration between the two.

[0052] 2. The control strategy of the present invention can be adjusted dynamically to generate optimal driving instructions according to real-time traffic conditions and vehicle status to improve fuel efficiency and the overall efficiency of traffic flow.

[0053] 3. The present invention proposes a hierarchical energy efficiency control framework that can optimize the behavior of intelligent connected vehicles and other surrounding intelligent connected vehicles at both the micro and macro levels, thereby improving the energy efficiency of the overall transportation network.

[0054] 4. The control strategy in the present invention can dynamically adapt to different traffic environments and conditions, optimize real-time decision-making, and improve the flexibility and efficiency of the system.

[0055] Other features and advantages of the present invention will be described in detail in the subsequent specific implementation mode; and the parts not involved in the present invention are the same as the prior art or can be implemented by using the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present invention but do not constitute a limitation of the present invention. In the accompanying drawings:

[0057] Figure 1 It is a flow chart of an adaptive pruning intelligent connected vehicle trajectory control method provided in a preferred embodiment of the present invention;

[0058] Figure 2is a flowchart of step S10 provided in a preferred embodiment of the present invention;

[0059] Figure 3 It is a structural block diagram of an adaptive pruning intelligent connected vehicle trajectory control system provided in a preferred embodiment of the present invention.

[0060] Description of Reference Numerals

[0061] 1Trajectory planning module 2Trajectory optimization module

[0062] 3Surrounding intelligent connected vehicle change suggestion module 4Execution module DETAILED DESCRIPTION

[0063] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0064] like Figure 1 As shown, the present invention provides an adaptive pruning intelligent connected vehicle trajectory control method based on Monte Carlo search, the method comprising:

[0065] Step S10, using a Monte Carlo tree search algorithm to generate an optimized trajectory plan for the target intelligent connected vehicle;

[0066] Step S20, pruning the action space of trajectory planning, reducing the decision space by eliminating unsafe or infeasible actions, and improving the operation efficiency of the algorithm;

[0067] Step S30, according to the trajectory planning of the target intelligent networked vehicle, proposes optimal speed and lane change suggestions based on the current traffic conditions to other intelligent networked vehicles around it, so as to reduce traffic congestion and improve the overall efficiency of traffic flow;

[0068] Step S40, combined with a low-level motion controller based on model predictive control, ensures that the intelligent connected vehicle drives smoothly in the motion space generated by trajectory planning, while optimizing fuel efficiency and improving driving comfort.

[0069] In a preferred embodiment of the present invention, step S10 includes the following steps:

[0070] Step S11, using optimal control according to the following formula to constrain the trajectory planning of the intelligent connected vehicle, ensuring that the trajectory planning process meets the requirements of safety, stability and fuel economy, and takes into account driving efficiency, energy optimization and safety constraints:

[0071]

[0072] It is subject to the dynamic constraints:

[0073] s x (k+1)=s x (k)+T s v(k)cos(θ(k))

[0074] s y (k+1)=s y (k)+T s v(k)sin(θ(k))

[0075] v(k+1)=v(k)+T s a(k)

[0076] θ(k+1)=θ(k)+T s ω(k)

[0077]

[0078] v min ≤v(τ)≤v max ,a min ≤a(τ)≤a max

[0079] Where k is a certain moment, τ is a certain moment in the control time domain, and s x (k) is the longitudinal position of the vehicle, s y (k) is the lateral position of the vehicle, v(k) is the vehicle speed, θ is the heading angle, a(k) is the vehicle acceleration, ω(k) is the vehicle steering acceleration, T s is the sampling time, is the fuel consumption rate, J is the other consumption of the trip, x(τ) is the trip distance, u(τ) is the formation time, S(τ) is the feasible area of ​​the road, S j (τ) is the set of positions occupied by road entities at time τ, is the set of entities around the vehicle, S safe (τ) is the safe driving area of ​​the vehicle at time τ.

[0080] Step S12, searching towards a better trajectory based on the Monte Carlo tree search algorithm, and setting the following reward function to evaluate the effects of different trajectory planning to guide the algorithm to search for a more optimized trajectory:

[0081]

[0082] in:

[0083]

[0084] Where: Δs is the longitudinal driving distance, is the fuel consumption of Δs, w1 is the fuel consumption weight, r c For collision reward, r v (τ) is close to the expected speed v des The reward is w2, and w3 is the acceleration / deceleration penalty weight.

[0085] In the above scheme, the present invention utilizes the above scheme to plan the optimal driving trajectory for the intelligent connected vehicle.

[0086] In a preferred embodiment of the present invention, the fuel consumption rate in step S11 is The road test data is fitted with a cubic polynomial using the following formula:

[0087]

[0088] In a preferred embodiment of the present invention, step S20 includes the following steps:

[0089] Step S21, dividing the vehicle behavior space into 14 combinations according to acceleration and steering wheel angle according to the following formula;

[0090]

[0091] Where: The first row of the matrix represents the acceleration (in m / s 2 ), the second row of the matrix represents the yaw angular velocity and is calculated in step S11;

[0092] Step S21, eliminating the unreasonable acceleration speed state by the following formula:

[0093]

[0094] Where: is the distance to the nearest object (vehicle / obstacle) in front of the intelligent connected vehicle in the same lane. is the object speed, w v3 ∈[0.8,1] is the linearly changing adaptive weight, and η2 is the time tolerance.

[0095] In a preferred embodiment of the present invention, step S30 includes the following steps:

[0096] Step S31, using the following formula to plan the optimal speed for other surrounding intelligent connected vehicles;

[0097]

[0098] Where: q∈{lt,sm,rt} represents left lane change, lane keeping and right lane change respectively. is the distance to the nearest road entity, 500 meters ahead at q th The average speed of all vehicles in the lane, d min is the minimum safety distance, T n is the headway parameter.

[0099] In a preferred embodiment of the present invention, the framework of the model predictive control based on which the low-level motion controller in step S40 is solved is solved by an optimized interior point method; and the motion controller based on the Monte Carlo tree search algorithm of the intelligent connected vehicle is implemented by the following formula:

[0100]

[0101] Where: w4 is the weighted acceleration for improving comfort, w5 is the penalty weight for trajectory deviation, s MCTS (τ) is the position of the intelligent connected vehicle obtained by the MCTS trajectory planner.

[0102] like Figure 3 As shown, the present invention also provides an adaptive pruning intelligent networked vehicle trajectory control system based on Monte Carlo search, the system comprising:

[0103] Trajectory planning module 1, used to generate optimized trajectory planning for intelligent connected vehicles using Monte Carlo tree search algorithm;

[0104] Trajectory optimization module 2 is used to prune the action space of trajectory planning, reduce the decision space by eliminating unsafe or infeasible actions, and improve the operation efficiency of the algorithm;

[0105] The surrounding intelligent connected vehicle change suggestion module 3 is used to propose optimal speed and lane change suggestions based on the current traffic conditions to other surrounding intelligent connected vehicles according to the trajectory planning of the target intelligent connected vehicle, so as to reduce traffic congestion and improve the overall efficiency of traffic flow;

[0106] The execution module 4 is used to combine the low-level motion controller based on model predictive control to ensure that the intelligent connected vehicle drives smoothly in the action space generated by trajectory planning, while optimizing fuel efficiency and improving driving comfort.

[0107] In summary, the adaptive pruning intelligent connected vehicle trajectory control method and system provided by the present invention have the following beneficial effects when used:

[0108] 1. This invention proposes a new control strategy for autonomous driving networked environment by combining Monte Carlo tree search algorithm (MCTS) algorithm and model predictive control (MPC) low-level motion controller, which can not only optimize the driving trajectory of intelligent networked vehicles, but also provide suggestions for other surrounding intelligent networked vehicles to achieve effective collaboration between the two.

[0109] 2. The control strategy of the present invention can be adjusted dynamically to generate optimal driving instructions according to real-time traffic conditions and vehicle status to improve fuel efficiency and the overall efficiency of traffic flow.

[0110] 3. The present invention proposes a hierarchical energy efficiency control framework that can optimize the behavior of intelligent connected vehicles and other surrounding intelligent connected vehicles at both the micro and macro levels, thereby improving the energy efficiency of the overall transportation network.

[0111] 4. The control strategy in the present invention can dynamically adapt to different traffic environments and conditions, optimize real-time decision-making, and improve the flexibility and efficiency of the system.

[0112] The preferred embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, a variety of simple modifications can be made to the technical solution of the present invention, and these simple modifications all belong to the protection scope of the present invention.

[0113] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present invention will not further describe various possible combinations.

[0114] In addition, various embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the present invention, they should also be regarded as the contents disclosed by the present invention.

Claims

1. An adaptive pruning intelligent connected vehicle trajectory control method, characterized in that: The method comprises: Step S10, using a Monte Carlo tree search algorithm to generate an optimized trajectory plan for the target intelligent connected vehicle; Step S20, pruning the action space of trajectory planning, reducing the decision space by eliminating unsafe or infeasible actions, and improving the operation efficiency of the algorithm; Step S30, according to the trajectory planning of the target intelligent networked vehicle, proposes optimal speed and lane change suggestions based on the current traffic conditions to other intelligent networked vehicles around it, so as to reduce traffic congestion and improve the overall efficiency of traffic flow; Step S40, combining a low-level motion controller based on model predictive control to ensure that the intelligent connected vehicle drives smoothly in the motion space generated by trajectory planning, while optimizing fuel efficiency and improving driving comfort; The step S10 comprises the following steps: Step S11, using optimal control according to the following formula to perform constraint control on the trajectory planning of the intelligent connected vehicle: At the same time, it is subject to the dynamic constraints: In the formula For a certain moment, To control a certain moment in the time domain, for The vehicle's steering acceleration at the moment, for The speed of the vehicle at the moment, for The longitudinal position of the vehicle at the moment, for The lateral position of the vehicle at the moment, for The vehicle speed at the moment, for The heading angle at the moment, for The vehicle acceleration at that moment, for The vehicle's steering acceleration at that moment, is the sampling time, is the fuel consumption rate, For other expenses during the trip, for The vehicle travel distance at the time, for The vehicle travel time, for The road feasible area at the moment, Road entity A collection of positions that are occupied at all times, is the set of entities around the vehicle, For vehicles in Always drive in safe areas; Step S12, searching towards a better trajectory based on the Monte Carlo tree search algorithm, and setting the following reward function to evaluate the effects of different trajectory planning to guide the algorithm to search for a more optimized trajectory: in: Where: Longitudinal distance travelled, for fuel consumption, is the fuel consumption weight, For collision rewards, To approach the desired speed The reward is weighted by , is the acceleration / deceleration penalty weight, for The speed of the vehicle at the moment The expected speed of the vehicle at time The absolute value of the difference.

2. The adaptive pruning intelligent connected vehicle trajectory control method according to claim 1 is characterized in that: The fuel consumption rate in step S11 The road test data is fitted with a cubic polynomial using the following formula: is the speed of the intelligent connected vehicle, is the acceleration of the intelligent connected vehicle:

3. The adaptive pruning intelligent connected vehicle trajectory control method according to claim 1, characterized in that: The step S20 comprises the following steps: Step S21, dividing the vehicle behavior space into 14 combinations according to acceleration and steering wheel angle according to the following formula; Where: The first row of the matrix represents acceleration, the unit is , the second row of the matrix represents the yaw angular velocity and is calculated in step S11; Step S21, eliminating the unreasonable acceleration speed state by the following formula: when hour; Where: The maximum average speed for intelligent connected vehicles to avoid collisions. The maximum safe speed for manually driven vehicles. is the distance to the nearest object in front of the intelligent connected vehicle in the same lane. The object is a vehicle or an obstacle. is the object speed, is a linearly changing adaptive weight, is the time tolerance, is the maximum value of acceleration, The minimum safe distance.

4. The adaptive pruning intelligent connected vehicle trajectory control method according to claim 1, characterized in that: The framework of the model predictive control based on which the low-level motion controller in step S40 is solved is solved by the optimized interior point method; and the motion controller based on the Monte Carlo tree search algorithm of the intelligent networked vehicle is implemented by the following formula: Where: To improve comfort, weighting is minimized to minimize acceleration. is the trajectory deviation penalty weight, The position of the intelligent connected vehicle obtained by the MCTS trajectory planner.

Images