Longitudinal speed control system and method for autonomous vehicle based on game theory

Abstract

The application discloses a kind of longitudinal speed control system and method of automatic driving vehicle based on game theory, the system includes: ego game strategy module, for constructing the first strategy set about ego vehicle longitudinal driving game, calculate the first game income corresponding to ego vehicle longitudinal driving game strategy;Artificial driving vehicle game strategy module, for constructing the second strategy set about artificial driving vehicle lane changing game, calculate the second game income corresponding to artificial driving vehicle lane changing game strategy;Game solution module, for according to the first strategy set, second strategy set, first game income and second game income, based on master-slave game strategy, ego optimal longitudinal driving trajectory planning is carried out;The application can comprehensively consider driver behavior characteristics, environmental constraints and multi-objective optimization, carry out the game optimization between ego and artificial driving vehicle, improve the decision robustness of mixed driving environment, effectively inhibit unreasonable cut-in behavior and reduce collision risk.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle subsystem joint control, in particular to an automatic driving vehicle longitudinal speed control system and method based on game theory. BACKGROUND

[0002] At present, with the rapid development of automatic driving technology, automatic driving vehicles have gradually integrated into the traffic system and shared road resources with manually driven vehicles, forming a mixed driving road environment. However, the existing intelligent driving technology still has deficiencies in dealing with interactions with manually driven vehicles.

[0003] On the one hand, traditional intelligent driving decision-making methods often lack interactivity and are difficult to dynamically adjust their own strategies according to the behavior of manually driven vehicles, which easily leads to poor traffic flow or safety hazards;

[0004] On the other hand, in the face of unreasonable behavior of manually driven vehicles such as sudden cutting in or aggressive driving, the existing system lacks flexibility and judgment in response, sometimes being too conservative and affecting driving efficiency, and sometimes being insufficient in response and endangering driving safety.

[0005] Therefore, for the problem of intelligent vehicle decision-making in a human-machine mixed driving environment, a more advanced and intelligent solution is needed. Game theory, as a discipline that studies the game and interaction between different participants, provides a new idea for solving such problems. By using a decision-making method based on game theory, intelligent vehicles can more effectively respond to the complex behavior of manually driven vehicles, achieve efficient and comfortable driving, and improve the operation efficiency and safety of the overall traffic system. SUMMARY

[0006] The purpose of the present application is to provide an automatic driving vehicle longitudinal speed control system and method based on game theory, thereby solving all or one of the above problems in the prior art.

[0007] To solve the above technical problems, the specific technical solutions of the present application are as follows:

[0008] On the one hand, the present application provides an automatic driving vehicle longitudinal speed control system based on game theory, comprising:

[0009] A self-vehicle game strategy module is used to construct a first strategy set for the self-vehicle longitudinal driving game. Based on the self-vehicle driving state information, the neighbor vehicle driving state information and the first strategy set, and considering the driving safety factor, the passenger comfort factor and the traffic efficiency factor, the first game benefit corresponding to the self-vehicle longitudinal driving game strategy is calculated.

[0010] The artificial driving vehicle game strategy module is configured to construct a second strategy set for a lane-changing game of the artificial driving vehicle; and calculate a second game payoff corresponding to a lane-changing game strategy of the artificial driving vehicle according to self-vehicle driving state information of the artificial driving vehicle, neighbor-vehicle driving state information of the artificial driving vehicle, and the second strategy set, and taking into account factors of position advantage, comfort, traffic efficiency, and different driving styles of different drivers.

[0011] The game solving module is configured to calculate an optimal strategy of the self-vehicle based on a master-slave game strategy according to the first strategy set, the second strategy set, the first game payoff, and the second game payoff; and perform optimal longitudinal driving trajectory planning of the self-vehicle according to the optimal strategy of the self-vehicle.

[0012] The environment perception module is configured to cooperate with the self-vehicle game strategy module, the artificial driving vehicle game strategy module, and the game solving module to acquire corresponding self-vehicle state information or neighbor-vehicle state information.

[0013] As an improved solution, the self-vehicle is an autonomous vehicle.

[0014] The self-vehicle game strategy module is further configured to establish a longitudinal driving model of the autonomous vehicle using a quintic spline curve; sample execution times and corresponding target longitudinal positions of the autonomous vehicle under different longitudinal driving game strategies at equal intervals based on the longitudinal driving model of the autonomous vehicle; and construct the first strategy set according to a plurality of the execution times and a plurality of the target longitudinal positions.

[0015] As an improved solution, the self-vehicle game strategy module is further configured to calculate a driving safety payoff of the artificial driving vehicle to the autonomous vehicle according to a plurality of lane-changing options of the artificial driving vehicle.

[0016] The self-vehicle game strategy module is further configured to calculate a comfort payoff of the autonomous vehicle using acceleration and jerk of the autonomous vehicle during a longitudinal lane-changing process.

[0017] The self-vehicle game strategy module is further configured to calculate a traffic efficiency payoff of the autonomous vehicle using a difference between a real-time vehicle speed of the autonomous vehicle and a target vehicle speed of the autonomous vehicle during the longitudinal lane-changing process.

[0018] As an improved solution, the self-vehicle game strategy module is further configured to take a sum of the driving safety payoff, the comfort payoff, and the traffic efficiency payoff as the first game payoff.

[0019] As an improved scheme, the manual driving vehicle game strategy module is further used to establish a lane changing model of the manual driving vehicle by using a quintic spline curve; based on the lane changing model of the manual driving vehicle, a plurality of lane changing times and a plurality of corresponding target longitudinal positions of the manual driving vehicle are sampled at equal intervals, and the second strategy set is constructed according to the plurality of lane changing times of the manual driving vehicle and the plurality of target longitudinal positions of the manual driving vehicle.

[0020] As an improved scheme, the manual driving vehicle game strategy module is further used to calculate a position advantage benefit of the manual driving vehicle according to a position safety between the manual driving vehicle and the automatic driving vehicle in different lane changing selection conditions of the manual driving vehicle at different driving positions;

[0021] The manual driving vehicle game strategy module is further used to calculate a comfort benefit of the manual driving vehicle by using jerk of the manual driving vehicle in the lane changing game process;

[0022] The manual driving vehicle game strategy module is further used to calculate a passing efficiency benefit of the manual driving vehicle by using a difference between a real-time vehicle speed of the manual driving vehicle and a target vehicle speed of the manual driving vehicle in the lane changing game process;

[0023] The manual driving vehicle game strategy module is further used to set a manual driving vehicle aggressiveness coefficient according to a difference in driving style of a driver, and take a weighted sum result of the position advantage benefit, the comfort benefit and the passing efficiency benefit with respect to the manual driving vehicle aggressiveness coefficient as the second game benefit.

[0024] As an improved scheme, the lane changing selection of the manual driving vehicle comprises:

[0025] Driving to the front of the automatic driving vehicle;

[0026] Driving to the rear of the automatic driving vehicle;

[0027] Continuing to follow the preceding vehicle without lane changing;

[0028] The driving position corresponding to the lane changing selection of the manual driving vehicle comprises:

[0029] A first driving position: the manual driving vehicle is close to the automatic driving vehicle and a preceding vehicle of the automatic driving vehicle, or the manual driving vehicle is close to a preceding vehicle of the manual driving vehicle;

[0030] The second driving position is that the longitudinal position of the manually driven vehicle is between the automatically driven vehicle and the front vehicle of the manually driven vehicle, and the manually driven vehicle keeps a safe distance from the front vehicle of the manually driven vehicle.

[0031] The third driving position is different from the first driving position and the second driving position.

[0032] As an improved scheme, the manually driven vehicle game strategy module is further configured to set the real-time aggressiveness coefficient of the manually driven vehicle by using the master-slave non-cooperative game theory; when the aggressiveness coefficient of the manually driven vehicle tends to zero, the manually driven vehicle game strategy module determines that the manually driven vehicle is a conservative vehicle; when the aggressiveness coefficient of the manually driven vehicle tends to one, the manually driven vehicle game strategy module determines that the manually driven vehicle is an aggressive vehicle; and when the aggressiveness coefficient of the manually driven vehicle is between zero and one, the manually driven vehicle game strategy module determines that the manually driven vehicle is an ordinary vehicle.

[0033] As an improved scheme, the game solving module is further configured to: solve the optimal strategy of the ego vehicle according to the master-slave game strategy based on the first game payoff, the second game payoff, the first strategy set and the second strategy set; perform ego vehicle optimal longitudinal driving trajectory planning according to the optimal strategy of the ego vehicle, and establish a multi-objective trajectory planning cost function; calculate the optimal target longitudinal trajectory corresponding to the optimal strategy of the ego vehicle and the optimal longitudinal vehicle speed corresponding to the optimal target longitudinal trajectory based on the multi-objective trajectory planning cost function; and perform longitudinal driving control of the automatically driven vehicle according to the optimal longitudinal vehicle speed.

[0034] In another aspect, the present application further provides an automatic driving vehicle longitudinal speed control method based on game theory, comprising the following steps:

[0035] The ego vehicle longitudinal driving game payoff calculation step:

[0036] The first strategy set of the ego vehicle longitudinal driving game is constructed; the first game payoff corresponding to the ego vehicle longitudinal driving game strategy is calculated based on the ego vehicle driving state information, the neighboring vehicle driving state information and the first strategy set, and considering the driving safety factor, the passenger comfort factor and the traffic efficiency factor;

[0037] The manually driven vehicle lane changing game payoff calculation step:

[0038] construct a second strategy set about the lane-changing game of the human-driven vehicle;calculate the second game income corresponding to the lane-changing game strategy of the human-driven vehicle according to the self-vehicle driving state information of the human-driven vehicle, the neighbor-vehicle driving state information of the human-driven vehicle and the second strategy set, and considering the position advantage, the comfort, the traffic efficiency factor and the different driving styles of different drivers;

[0039] comprehensive game solving steps:

[0040] calculate the optimal strategy of the self-vehicle based on the master-slave game strategy according to the first strategy set, the second strategy set, the first game income and the second game income;perform the optimal longitudinal driving trajectory planning of the self-vehicle according to the optimal strategy of the self-vehicle.

[0041] The beneficial effects of the technical scheme of the application are:

[0042] 1. The automatic driving vehicle longitudinal speed control system based on game theory can establish a multi-dimensional game strategy model of human-driven vehicles and automatic driving vehicles through the cooperation of system modules, comprehensively consider the driver behavior characteristics, environmental constraints and multi-objective optimization, and improve the decision robustness of mixed driving environment;The system can dynamically evaluate the position advantage and strategy income, effectively suppress unreasonable cutting behavior and reduce the risk of collision, while balancing comfort and traffic efficiency;Real-time strategy optimization is realized based on the master-slave game method, which ensures that the self-vehicle quickly responds in complex interactive scenarios, significantly enhances the environmental adaptability, and finally achieves the driving goal of safe, efficient and comfortable driving of automatic driving vehicles, which is significantly better than the traditional static decision scheme.

[0043] 2. The automatic driving vehicle longitudinal speed control method based on game theory can orderly call the system modules, thereby realizing the system logic of the automatic driving vehicle longitudinal speed control system based on game theory. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the specific embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0045] Figure 1 is the schematic diagram of the principle architecture of the automatic driving vehicle longitudinal speed control system based on game theory according to embodiment 1 of the present application;

[0046] Figure 2is a strategy set diagram when the human-driven vehicle lane-changing time is t in the automatic driving vehicle longitudinal speed control system based on game theory in embodiment 1 of the present application i

[0047] Figure 3 is a driving trajectory diagram of the vehicle in the simulation process of the automatic driving vehicle longitudinal speed control system based on game theory in embodiment 1 of the present application

[0048] Figure 4 is an estimation result diagram of the vehicle speed and driving style in the simulation process of the automatic driving vehicle longitudinal speed control system based on game theory in embodiment 1 of the present application

[0049] Figure 5 is a flow diagram of the automatic driving vehicle longitudinal speed control method based on game theory in embodiment 2 of the present application. DETAILED DESCRIPTION

[0050] The preferred embodiments of the present application will be described in detail below with reference to the accompanying drawings, so that the advantages and features of the present application can be more easily understood by those skilled in the art, and the scope of protection of the present application can be more clearly defined.

[0051] In the description of the present application, it should be noted that the embodiments described in the present application are part of the embodiments of the present application, not all the embodiments; all other embodiments obtained by those skilled in the art without creative labor based on the embodiments in the present application, belong to the scope of protection of the present application.

[0052] The terms "first", "second", and the like in the specification and claims herein and the above-described drawings are used to distinguish similar objects, and do not necessarily describe a particular order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, device, product or equipment including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or equipment.

[0053] Embodiment 1, the present embodiment provides an automatic driving vehicle longitudinal speed control system based on game theory, as shown in Figures 1-4 , comprising:

[0054] (1) an environment perception module for obtaining self-driving state information and adjacent vehicle driving state information;

[0055] ​Specifically, the ego vehicle state information includes: vehicle position and speed of the ego vehicle (EV); wherein the ego vehicle is an autonomous vehicle;

[0056] Specifically, the neighboring vehicle state information includes: vehicle position and speed of the human-driven vehicle (HV) about to change lanes, the front vehicle (TP) in the lane where the ego vehicle is located, and the front vehicle (PV) in the lane where the human-driven vehicle is located.

[0057] (2) The ego vehicle game strategy module is configured to calculate the game benefit of the ego vehicle according to the ego vehicle driving state information and the neighboring vehicle driving state information;

[0058] Specifically, the ego vehicle game strategy module establishes an ego vehicle longitudinal driving model; based on the ego vehicle longitudinal driving model, the game strategy execution time of the EV within a certain range and the target longitudinal position of the EV are sampled to construct a two-dimensional strategy set S EV ; based on the two-dimensional strategy set S EV , and considering the driving safety, passenger comfort and traffic efficiency factors, the game benefit corresponding to each strategy and the final total game benefit are calculated;

[0059] Specifically, the above process performed by the ego vehicle game strategy module includes:

[0060] (2.1) The ego vehicle game strategy module establishes a longitudinal model of the autonomous vehicle using a quintic spline curve, wherein the longitudinal position, speed and acceleration of the vehicle are represented as follows:

[0061] ;

[0062] ;

[0063] ;

[0064] wherein c i is a polynomial coefficient, c0 is the longitudinal position of the vehicle at time t0, c1 is the speed of the vehicle at time t0, 2c2 is the acceleration of the vehicle at time t0, 、 and are the coefficients to be solved in the quintic spline curve.

[0065] (2.2) The ego vehicle game strategy module collects different game strategy execution times t f and corresponding target longitudinal positions x f of the autonomous vehicle as the first strategy set of the autonomous vehicle; the first strategy set is used to represent the response of the EV to the lane changing behavior of the HV, denoted as S EV ; the number of strategies in S EV is N EV1 ×NEV2 , S EV is expressed as follows:

[0066] .

[0067] (2.3) The self-game strategy module calculates the game income corresponding to the strategy and the final total game income, as follows:

[0068] (2.3.1) For the autonomous vehicle EV, the human-driven vehicle HV has two different choices when changing lanes, including: driving before the autonomous vehicle EV and driving after the autonomous vehicle EV; these two different choices will bring different driving safety income to the autonomous vehicle EV, and the self-game strategy module measures the driving safety income between EV and HV as the headway T h , and the specific calculation formula is as follows:

[0069] ;

[0070] ;

[0071] where w safe1 is the weight coefficient, x HV is the longitudinal position of HV after the game decision, v HV is the vehicle speed of HV after the game decision, x EV is the longitudinal position of EV for game decision, and v EV is the vehicle speed of EV for game decision.

[0072] (2.3.2) Since there may also be a front vehicle TP on the lane where EV is located, the self-game strategy module calculates the driving safety income between EV and TP , which is specifically expressed as follows:

[0073] ;

[0074] where w safe2 is the weight coefficient, and x TP is the longitudinal position of TP after the game decision of EV.

[0075] (2.3.3) The self-game strategy module uses the acceleration and jerk in the process of EV executing the longitudinal strategy to represent the comfort income of EV , , which is specifically as follows:

[0076] ;

[0077] where w comi is the weight coefficient. ​Let time t be the acceleration The calculation function, Let time t be the acceleration The calculation function.

[0078] (2.3.4) The self-vehicle game strategy module uses the difference between the real-time speed of the EV and the target speed during the EV's execution of the longitudinal strategy to represent the corresponding traffic efficiency gain. The details are as follows:

[0079] ;

[0080] Among them, w eff v is the weighting coefficient. d,EV The target speed for the EV;

[0081] (2.3.5) The autonomous vehicle game strategy module adds up the driving safety benefits, comfort benefits, and traffic efficiency benefits to obtain the total game benefits of the autonomous vehicle EV. (i.e., the payoff of the first game), as detailed below:

[0082] .

[0083] (3) The game strategy module for manually driven vehicles is used to calculate the game payoff of manually driven vehicles based on their own driving status information and the driving status information of neighboring vehicles.

[0084] Specifically, the game strategy module for the manually driven vehicle establishes a lane-changing model for the manually driven vehicle. Based on this model, it samples the lane-changing time and target longitudinal position of the vehicle (HV) within a certain range at equal intervals, and constructs a corresponding two-dimensional strategy set S. HV Based on S HV Taking into account factors such as location advantage, comfort, traffic efficiency, and the different driving styles of different drivers, the game payoff for each strategy and the final total game payoff are calculated.

[0085] Specifically, the process executed by the game strategy module of the manually driven vehicle includes:

[0086] (3.1) The game strategy module for manually driven vehicles uses quintic spline curves to establish a lane-changing model for manually driven vehicles, describing the longitudinal and lateral movements of the vehicles, as specifically expressed below:

[0087] ;

[0088] Where x is the longitudinal coordinate of the manually driven vehicle's HV, y is the lateral coordinate of the manually driven vehicle's HV, and a i (i=0,1,2,…5), b j(j = 0, 1, 2, … 5) are polynomial coefficients of the lane-changing trajectory.

[0089] (3.2) The game strategy module of the human-driven vehicle, which samples the lane-changing time and the corresponding target longitudinal position of the human-driven vehicle HV in a certain range at equal intervals, and constructs a second strategy set S HV , as follows:

[0090] (3.2.1) The driving time range corresponding to the lane-changing process is expressed as follows:

[0091] ;

[0092] where T min is the minimum lane-changing time acceptable to the human driver, T max is the maximum lane-changing time acceptable to the human driver, t0 is the start time of the lane-changing process, and t f is the end time of the lane-changing process;

[0093] (3.2.2) The lane-changing time is sampled at equal intervals between T min and T max , and is expressed as follows:

[0094] ;

[0095] where ΔT is the sampling time interval, N HV1 is the number of lane-changing time samples, t m is the mth sampled lane-changing time, and the trajectory can be optimized based on different lane-changing times;

[0096] (3.2.3) The target lateral position of the lane-changing of the human-driven vehicle HV is the target lane center line, and the value range of the target longitudinal position x f of HV is set as follows, considering safety and efficiency and other factors:

[0097] ;

[0098] where x TP is the longitudinal position of TP after lane-changing, x PV is the longitudinal position of PV after lane-changing, x0 is the longitudinal position of HV at the current time, X safe is the safety distance, X safe = v f *T h,min , v f is the vehicle speed of EV at the end of lane-changing, and T h,min is the minimum vehicle headway to ensure safety;

[0099] (3.2.4) x fEqual-interval sampling is performed, and the number of samples is N HV2， The specific representation is as follows:

[0100] ;

[0101] (3.2.5) As Figure 2 , Figure 2 the horizontal coordinate is different lane changing time t f , and the vertical coordinate is the target longitudinal position x f corresponding to different lane changing time, based on which, considering the importance of lane changing time and target longitudinal position in the lane changing process, the artificial driving vehicle game strategy module selects different lane changing time t f and the corresponding target longitudinal position x f as the second strategy set of the artificial driving vehicle HV, denoted as S HV , S HV characterizes the response of drivers of different styles to the behavior of the ego vehicle, and the number of strategies of S HV is N HV1 × N HV2 , and S HV is specifically represented as follows:

[0102] .

[0103] (3.3) The artificial driving vehicle game strategy module calculates the game income corresponding to each strategy and the final total game income, which is specifically as follows:

[0104] (3.3.1) When the artificial driving vehicle HV is lane changing, the lane changing options include:

[0105] A, driving to the front of the autonomous vehicle EV;

[0106] B, driving to the rear of the autonomous vehicle EV;

[0107] C, continuing to follow the preceding vehicle PV without lane changing.

[0108] (3.3.2) According to the intention of artificial driving, lane changing will be selected under the following conditions:

[0109] A, when HV is too close to EV and TP or too close to PV, HV selects to continue following;

[0110] B, when the HV longitudinal position is between EV and TP, and maintains a safe distance from other vehicles, HV selects to merge into the front of EV;

[0111] C, in other cases, HV selects to merge into the rear of EV;

[0112] The regions Region corresponding to the above different lane changing decisions are as follows:

[0113] ;

[0114] wherein, x TP is the longitudinal position of the TP when the manual driving vehicle ends the lane change; EV is the longitudinal position of the EV when the manual driving vehicle ends the lane change; PV is the longitudinal position of the PV when the manual driving vehicle ends the lane change; HV is the longitudinal position of the HV when the manual driving vehicle ends the lane change.

[0115] (3.3.3) Different lane change selections are made when the longitudinal position of the HV is in different Regions:

[0116] A. When the longitudinal position of the HV is in Region 1 and Region 2, the HV continues to follow the vehicle;

[0117] B. When the longitudinal position of the HV is in Region 3, the HV will cut into the front of the EV;

[0118] C. When the longitudinal position of the HV is in Region 4, the HV will cut into the rear of the EV.

[0119] (3.3.4) Under different lane change selections, the corresponding position advantage benefits are different. When the HV selects to drive to the front of the EV, the HV will be more expected to drive to the middle position between the TP and the EV and maintain a safe distance with the PV. When the HV selects to drive to the rear of the EV, the HV will be expected to maintain a safe distance with the EV and the PV. The position advantage benefit of the HV is calculated by the manual driving vehicle game strategy module as follows:

[0120] ;

[0121] wherein, W safe is the position benefit weight coefficient, and W is a constant set; wherein, when, ; when, ; when, .

[0122] (3.3.5) The manual driving vehicle game strategy module uses the jerk in the HV game process to represent the comfort benefit of the HV , which is specifically as follows:

[0123] ;

[0124] wherein, W com is the comfort benefit weight coefficient. ​

[0125] (3.3.6) The game strategy module for manually driven vehicles uses the difference between the real-time vehicle speed and the target vehicle speed during the HV game to represent the traffic efficiency gain. The details are as follows:

[0126] ;

[0127] Among them, W eff v is the comfort benefit weighting coefficient. d,HV The target speed for HV.

[0128] (3.3.7) The game strategy module for manually driven vehicles considers the different driving styles of different drivers and calculates the total game payoff function of HV. (i.e., the payoff in the second game) is expressed as:

[0129] ;

[0130] Where ρ is the aggressiveness coefficient of manually driven vehicles;

[0131] When ρ approaches 0 (e.g., ρ is within the first interval [0.01~0.3], the first interval is set according to the actual situation), HV is a conservative vehicle. Conservative vehicles pay more attention to driving safety and often choose to decelerate and merge into the target lane, and keep a large safety distance from the vehicles in front and behind in the target lane.

[0132] When ρ approaches 1 (e.g., ρ is within the second interval (0.7~0.99], the second interval is set according to the actual situation), HV is an aggressive vehicle. Aggressive vehicles often choose to accelerate and merge into the target lane in order to gain a greater positional advantage and traffic efficiency.

[0133] When ρ is between 0 and 1 (e.g., ρ is in the third interval [0.3~0.7], the third interval is set according to the actual situation), HV is a normal vehicle, which is between a conservative vehicle and an aggressive vehicle, and prefers driving comfort.

[0134] (3.3.8) The game strategy module for manually driven vehicles is based on the master-slave non-cooperative game theory to set up an online driving style identification method, namely, a real-time update algorithm for ρ, as follows:

[0135] ;

[0136] in, This represents the estimated driving style of a human driver; s EV The game strategy adopted by EV; s HV,observe The game strategy adopted by the human driver, obtained by the EV through sensors or vehicle-to-vehicle communication; represents the game strategy adopted by the human driver at the last time step; the artificial driving vehicle game strategy module solves the above inequality in real time, and then obtains the estimated artificial driving vehicle aggressive driving style coefficient p; it should be noted that at the initial time, the driving style coefficient is set as p = 0.5 by default; at each time step thereafter, the artificial driving vehicle game strategy module solves the inequality in real time according to the observed actual behavior of the HV, and realizes real-time updating of p.

[0137] (4) A game solving module, configured to solve the optimal strategy of the ego vehicle based on the master-slave game strategy according to the first strategy set of the EV, the second strategy set of the HV, the first game payoff of the EV, and the second game payoff of the HV; performing longitudinal planning of the ego vehicle according to the solved optimal strategy of the ego vehicle, and solving the corresponding speed of the ego vehicle within the planning execution time, specifically as follows:

[0138] (4.1) The game solving module solves the optimal strategy of the ego vehicle based on the first game payoff, the second game payoff, S EV , and S HV , according to the master-slave game strategy, specifically as follows:

[0139] (4.1.1) Since TP and PV are both artificial driving vehicles, it is assumed that TP and PV will not participate in the game of the rear vehicle, and TP and PV are regarded as moving obstacles; finally, considering the interaction between the EV and the HV, the game solving module solves the leader game payoff and describes it as:

[0140] ;

[0141] wherein s EV is the strategy of the EV; s HV is the strategy of the HV; J(s EV , s HV ) represents the maximum game payoff that the leader EV can obtain when the EV and the HV adopt s EV and s HV strategies respectively; J EV is the first game payoff.

[0142] wherein when the HV selects s HV strategy, the future driving trajectory of the HV is predicted in real time according to the current position and speed of the HV by using the spline curve method; when the EV selects s EV strategy, the longitudinal position, speed and acceleration of the EV are calculated by combining the spline curve and the linear quadratic programming method; based on the future driving trajectory of the HV, the longitudinal position, speed and acceleration of the EV, the maximum game payoff J(s EV , s HV ) corresponding to the selected strategy set is calculated.

[0143] (4.1.2) Considering the interaction between EV and HV, the optimal strategy of EV is calculated based on the master-slave game strategy and J(s EV , s HV ) as follows:

[0144] ;

[0145] ;

[0146] wherein, is the optimal strategy of HV, is the optimal strategy of EV, i.e. the equilibrium solution; is the optimal strategy set of HV.

[0147] (4.2) The game solution module performs longitudinal planning of the ego vehicle according to the optimal strategy obtained by the solution, and solves the optimal speed of the ego vehicle within the execution time of the game strategy;

[0148] Specifically, the game solution module establishes a multi-objective trajectory planning cost function based on the trajectory model of the quintic spline curve, and solves the corresponding optimal target longitudinal trajectory within the execution time t of the current game strategy based on the multi-objective trajectory planning cost function.

[0149] Specifically, in the quintic spline curve, the polynomial coefficients c i determine the specific trajectory of the vehicle, c0, c1, 2c2 are the known state parameters of the vehicle at t0 (same as 2.1); c3, c4 and c5 are unknown quantities, which are to be solved coefficients and need to be solved by establishing an objective function; when the EV performs longitudinal trajectory planning, it needs to construct a longitudinal trajectory planning objective function according to the comfort, efficiency and position requirements, and the specific calculation process is as follows:

[0150] (4.2.1) The game solution module uses the acceleration and jerk of the EV in the process of executing the longitudinal strategy to represent the comfort benefit of the EV , as follows:

[0151] ;

[0152] wherein, w coi is the comfort benefit weight coefficient.

[0153] (4.2.2) The game solution module uses the difference between the real-time speed of the EV and the target speed of the EV in the process of executing the longitudinal strategy, as well as the difference between the speed of the EV at the end of the longitudinal planning and the target speed of the EV, to represent the traffic efficiency benefit , as follows:

[0154] ;

[0155] wherein, wefi is a passing efficiency weight coefficient, v d,EV is a target vehicle speed.

[0156] (4.2.3) The game solving module considers whether the longitudinal position of the EV at the planning endpoint can maintain a safe distance with the HV and the TP, and represents the position benefit based on this safe distance factor , as follows:

[0157] ;

[0158] wherein w posi is a weight coefficient, x TP is the longitudinal position of the TP at the end of the game, x HV is the longitudinal position of the HV at the end of the game.

[0159] (4.2.4) The game solving module combines the above benefit function and considers the position, speed and acceleration constraints to obtain a multi-objective trajectory planning cost function , as follows:

[0160] ;

[0161] .

[0162] (4.2.5) The game solving module takes the longitudinal position, speed and acceleration of the longitudinal planning endpoint position as optimization quantities based on the multi-objective trajectory planning cost function, and obtains the optimal x(t f ), v(t f ) and a(t f ) through a QP algorithm; based on the following transformation, the longitudinal trajectory polynomial coefficient c i is obtained, and then the optimal target longitudinal trajectory corresponding to the current t f is obtained based on c i , and the formula transformation is as follows:

[0163] .

[0164] It should be noted that the system finally controls the vehicle according to the longitudinal vehicle speed of the optimal target longitudinal trajectory obtained, and controls the above modules to operate in a logical order in a loop, continuously updates the vehicle state information and the longitudinal vehicle speed of the optimal target longitudinal trajectory, and realizes continuous vehicle control until the vehicle is in a stopped state.

[0165] As a preferred embodiment of the present application, in order to verify whether the present application can improve the response capability of an autonomous vehicle when facing unreasonable cut-in behavior of a manually driven vehicle, a Matlab simulation program is built for simulation test as follows:

[0166] (i) The driving conditions are set as follows:

[0167] Simulation vehicles: a front vehicle TP on the left lane, an automatic driving rear vehicle EV, a front vehicle PV on the right lane, and a human driving lane-changing vehicle HV;

[0168] Initial longitudinal positions: TP is 100 m, PV is 140 m, EV is 40 m, and HV is 50 m; during the test, TP and PV travel on their respective lanes at constant speeds of 12 m / s and 15 m / s, respectively; at the beginning of the simulation, EV travels in a following mode at a target speed of 15 m / s; HV travels on the right lane and is ready to merge into the left lane due to the slow speed of PV; the driving style of HV is set to be moderate.

[0169] (ii) The final simulation results are shown in Figure 3 and Figure 4 ( Figure 3 The horizontal coordinate is the longitudinal position of the vehicle, the horizontal coordinate is the lateral position of the vehicle, the dashed line in the middle is a lane line that divides the left lane and the right lane, and the continuous curved line represents the driving trajectory of HV); based on the simulation results, it can be seen that the ego vehicle can actively respond to the cutting behavior of the moderate aggressive human driving vehicle by accelerating to stop the human driving vehicle from cutting in front of the ego vehicle, ensuring the driving efficiency of the ego vehicle and improving the intelligence of the automatic driving.

[0170] It should be noted that all the above examples are only for the purpose of explaining the present application and cannot limit the protection scope of the present application.

[0171] Embodiment 2, based on the same inventive concept as the automatic driving vehicle longitudinal speed control system based on game theory described in Embodiment 1, provides an automatic driving vehicle longitudinal speed control method based on game theory, as shown in Figure 5 , including the following steps:

[0172] S100, a self-vehicle longitudinal driving game benefit calculation step, including:

[0173] constructing a first strategy set for the longitudinal driving game of the ego vehicle; based on the ego vehicle driving state information, the neighbor vehicle driving state information, and the first strategy set, and considering the driving safety factor, the passenger comfort factor, and the traffic efficiency factor, calculating the first game benefit corresponding to the longitudinal driving game strategy of the ego vehicle;

[0174] S200, a human driving vehicle lane-changing game benefit calculation step, including:

[0175] construct a second strategy set for the lane-changing game of the human-driven vehicle; calculate a second game payoff corresponding to the lane-changing game strategy of the human-driven vehicle according to the self-vehicle driving state information of the human-driven vehicle, the neighboring vehicle driving state information of the human-driven vehicle, and the second strategy set, and taking into account the positional advantage, comfort, traffic efficiency factors, and different driving styles of different drivers;

[0176] S300, a comprehensive game solving step, comprising:

[0177] According to the first strategy set, the second strategy set, the first game payoff, and the second game payoff, calculate the optimal strategy of the self-vehicle based on the master-slave game strategy; and perform optimal longitudinal driving trajectory planning of the self-vehicle according to the optimal strategy of the self-vehicle.

[0178] Unlike the prior art, the automatic driving vehicle longitudinal speed control system and method based on game theory can establish a multi-dimensional game strategy model of human-driven vehicles and automatic driving vehicles, comprehensively consider driver behavior characteristics, environmental constraints, and multi-objective optimization, and improve the decision robustness of mixed driving environments. The system can dynamically evaluate the positional advantage and strategy payoff, effectively suppress unreasonable cut-in behavior and reduce collision risk, while balancing comfort and traffic efficiency. Based on the master-slave game method, real-time strategy optimization is realized to ensure that the self-vehicle quickly responds in complex interactive scenarios, significantly enhances the environmental adaptability, and ultimately achieves the driving goal of safe, efficient, and comfortable driving of the automatic driving vehicle, which is significantly better than the traditional static decision scheme.

[0179] It should be understood that in various embodiments herein, the size of the sequence number of each process described above does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments herein.

[0180] It should also be understood that in the embodiments herein, the term "and / or" is only a description of the association relationship between the associated objects, which means that there can be three relationships. For example, A and / or B can represent three cases: A exists alone, A and B exist together, and B exists alone. In addition, the character " / " in this paper generally represents that the front and rear associated objects are a "or" relationship.

[0181] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific logical process of the above-described method can refer to the corresponding working process of the system, device and unit in the foregoing method embodiments, which will not be described here.

[0182] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific logical process of the above-described method can refer to the corresponding working process of the system, device and unit in the foregoing method embodiments, which will not be described here.

[0183] In several embodiments provided herein, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the above-described device embodiments are merely illustrative, for example, the division of the units is merely a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be omitted or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can also be electrical, mechanical or other forms of connection.

[0184] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0185] In addition, each functional unit in each embodiment herein can be integrated in one processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0186] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions herein or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments herein. The aforementioned storage medium includes various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0187] The above description is only an embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields, which is made by using the content of the specification and drawings of the present application, is also included in the patent protection scope of the present application.

Claims

1. A longitudinal speed control system for autonomous vehicles based on game theory, characterized in that, The system comprises: a self-vehicle game strategy module, configured to construct a first strategy set for a self-vehicle longitudinal driving game; based on the self-vehicle driving state information, the adjacent vehicle driving state information, and the first strategy set, and considering the driving safety factor, the passenger comfort factor, and the traffic efficiency factor, to calculate a first game benefit corresponding to the self-vehicle longitudinal driving game strategy; a human-driven vehicle game strategy module, configured to construct a second strategy set for a human-driven vehicle lane-changing game; according to the self-vehicle driving state information of the human-driven vehicle, the adjacent vehicle driving state information of the human-driven vehicle, and the second strategy set, and considering the position advantage, the comfort, the traffic efficiency factor, and the different driving styles of different drivers, to calculate a second game benefit corresponding to the human-driven vehicle lane-changing game strategy; a game solving module, configured to calculate an optimal strategy of the self-vehicle based on the first strategy set, the second strategy set, the first game benefit, and the second game benefit, and based on the master-slave game strategy; to perform optimal longitudinal driving trajectory planning of the self-vehicle according to the optimal strategy of the self-vehicle; an environment perception module, configured to cooperate with the self-vehicle game strategy module, the human-driven vehicle game strategy module, and the game solving module to acquire corresponding self-vehicle state information or adjacent vehicle state information; the self-vehicle is an autonomous vehicle; the self-vehicle game strategy module is further configured to establish a longitudinal driving model of the autonomous vehicle by using a quintic spline curve; to sample the execution time and the corresponding target longitudinal position of the autonomous vehicle under different longitudinal driving game strategies at equal intervals based on the longitudinal driving model of the autonomous vehicle; and to construct the first strategy set according to a plurality of the execution time and a plurality of the target longitudinal position; the self-vehicle game strategy module is further configured to calculate the driving safety benefit of the human-driven vehicle to the autonomous vehicle according to a plurality of lane-changing options of the human-driven vehicle; the calculation of the driving safety benefit of the human-driven vehicle to the autonomous vehicle further comprises: The ego game strategy module measures the driving safety benefit between EV and HV with the headway T h measures the driving safety benefit between EV and HV The calculation formula is as follows: ； ； where EV is the ego vehicle, HV is the human-driven vehicle, w safe1 is the weight coefficient, x HV is the longitudinal position of the HV after the game decision, v HV is the speed of the HV after the game decision, x EV is the longitudinal position of the EV that makes the game decision, v EV is the speed of the EV that makes the game decision; when the longitudinal position of the HV is in Region 3, the HV will cut in front of the EV; when the longitudinal position of the HV is in Region 4, the HV will cut in behind the EV; the self-vehicle game strategy module is further configured to calculate the comfort benefit of the autonomous vehicle according to the acceleration and the jerk of the autonomous vehicle during the longitudinal lane-changing process of the autonomous vehicle; the self-vehicle game strategy module is further configured to calculate the traffic efficiency benefit of the autonomous vehicle according to the difference between the real-time vehicle speed of the autonomous vehicle and the target vehicle speed of the autonomous vehicle during the longitudinal lane-changing process of the autonomous vehicle; the human-driven vehicle game strategy module is further configured to establish a lane-changing model of the human-driven vehicle by using a quintic spline curve; to sample a plurality of lane-changing times and a plurality of target longitudinal positions of the human-driven vehicle at equal intervals based on the lane-changing model of the human-driven vehicle; and to construct the second strategy set according to the plurality of the lane-changing times of the human-driven vehicle and the plurality of target longitudinal positions of the human-driven vehicle.

2. The longitudinal speed control system of the autonomous vehicle based on the game theory according to claim 1, wherein: The self-driving vehicle game strategy module is further configured to: take a sum of the driving safety benefit, the comfort benefit, and the traffic efficiency benefit as the first game benefit. 3.The game theory based automatic driving vehicle longitudinal speed control system according to claim 1, wherein: The human-driven vehicle game strategy module is further configured to: calculate a position advantage benefit of the human-driven vehicle according to a position safety between the human-driven vehicle and the automatic driving vehicle in different lane-changing selection situations of the human-driven vehicle at different driving positions; The human-driven vehicle game strategy module is further configured to: calculate a comfort benefit of the human-driven vehicle by using jerk of the human-driven vehicle in the lane-changing game process; The human-driven vehicle game strategy module is further configured to: calculate a traffic efficiency benefit of the human-driven vehicle by using a difference between a real-time vehicle speed of the human-driven vehicle and a target vehicle speed of the human-driven vehicle in the lane-changing game process; The human-driven vehicle game strategy module is further configured to: set a human-driven vehicle aggressiveness coefficient according to a difference in driving styles of drivers, and take a weighted sum of the position advantage benefit, the comfort benefit, and the traffic efficiency benefit with respect to the human-driven vehicle aggressiveness coefficient as the second game benefit. 4.The game theory based automatic driving vehicle longitudinal speed control system according to claim 3, wherein: The lane-changing selection of the human-driven vehicle includes: driving to the front of the automatic driving vehicle; driving to the back of the automatic driving vehicle; continuing to follow the front vehicle without lane-changing; The driving position corresponding to the lane-changing selection of the human-driven vehicle includes: a first driving position: the human-driven vehicle is close to the automatic driving vehicle and a front vehicle of the automatic driving vehicle, or the human-driven vehicle is close to a front vehicle of the human-driven vehicle; a second driving position: a longitudinal position of the human-driven vehicle is between the automatic driving vehicle and the front vehicle of the automatic driving vehicle, and the human-driven vehicle maintains a safe distance from the front vehicle of the human-driven vehicle; a third driving position: a driving position other than the first driving position and the second driving position. 5.The game theory based automatic driving vehicle longitudinal speed control system according to claim 3, wherein: The human-driven vehicle game strategy module is further configured to: set the real-time human-driven vehicle aggressiveness coefficient by using a master-slave non-cooperative game theory; when the human-driven vehicle aggressiveness coefficient is in a first interval, the human-driven vehicle game strategy module determines that the human-driven vehicle is a conservative vehicle; when the human-driven vehicle aggressiveness coefficient is in a second interval, the human-driven vehicle game strategy module determines that the human-driven vehicle is an aggressive vehicle; and when the human-driven vehicle aggressiveness coefficient is in a third interval, the human-driven vehicle game strategy module determines that the human-driven vehicle is an ordinary vehicle. The first interval corresponds to a value of zero, the second interval corresponds to a value of one, and the third interval is between the first and second intervals. 6.The game theory based longitudinal speed control system for autonomous vehicle according to claim 1, wherein: The game solving module is further configured to: based on the first game payoff, the second game payoff, the first strategy set and the second strategy set, solve the optimal strategy of the ego vehicle according to a master-slave game strategy; perform optimal longitudinal driving trajectory planning of the ego vehicle according to the optimal strategy of the ego vehicle, and establish a multi-objective trajectory planning cost function; calculate an optimal target longitudinal trajectory corresponding to the optimal strategy of the ego vehicle and an optimal longitudinal vehicle speed corresponding to the optimal target longitudinal trajectory based on the multi-objective trajectory planning cost function; and perform longitudinal driving control of the autonomous vehicle according to the optimal longitudinal vehicle speed.

7. A game theory based method for longitudinal speed control of an autonomous vehicle based on the game theory based longitudinal speed control system of any one of claims 1-6, characterized in that, The game theory based longitudinal speed control method for autonomous vehicle comprises the following steps: The ego vehicle longitudinal driving game payoff calculation step: construct a first strategy set for the ego vehicle longitudinal driving game; based on the ego vehicle driving state information, the neighboring vehicle driving state information and the first strategy set, and considering the driving safety factor, the passenger comfort factor and the traffic efficiency factor, calculate the first game payoff corresponding to the ego vehicle longitudinal driving game strategy; The human-driven vehicle lane changing game payoff calculation step: construct a second strategy set for the human-driven vehicle lane changing game; based on the ego vehicle driving state information of the human-driven vehicle, the neighboring vehicle driving state information of the human-driven vehicle and the second strategy set, and considering the position advantage, the comfort, the traffic efficiency factor and the different driving styles of different drivers, calculate the second game payoff corresponding to the human-driven vehicle lane changing game strategy; The comprehensive game solving step: based on the first strategy set, the second strategy set, the first game payoff and the second game payoff, calculate the optimal strategy of the ego vehicle based on a master-slave game strategy; and perform optimal longitudinal driving trajectory planning of the ego vehicle according to the optimal strategy of the ego vehicle.

Images