An autonomous driving testing system under a strong game-theoretic interactive environment

Abstract

This invention relates to an autonomous driving testing system in a strongly game-theoretic interactive environment, comprising: a traffic simulator for simulating a traffic system; an interactive game-theoretic decision-maker that takes road traffic information output by the traffic simulator as input, embeds a game-theoretic decision-making algorithm based on logical rules, classifies vehicle operation modes into five categories: cruising, lane changing, returning to the lane, traffic lights, and conflict zones, constructs a decision-making mechanism for each operation mode, determines the operation mode for the next simulation step, and outputs the corresponding vehicle state; an interactive game-theoretic controller that outputs control quantities for controlling the lateral and longitudinal behavior of the autonomous vehicle based on the vehicle state output by the interactive game-theoretic decision-maker; and a vehicle dynamics model that simulates the real response of the vehicle's mechanical system to the control quantities. Compared with existing technologies, this invention has the advantages of clearly depicting vehicle decision-making behavior, meeting real-time simulation requirements, and providing high reliability in simulation testing.

Description

Technical Field

[0001] This invention relates to the field of autonomous vehicle simulation testing, and in particular to an autonomous driving testing system under a strong game-theoretic interactive environment. Background Technology

[0002] With the rise of a new round of technological revolution represented by mobile internet, big data, and cloud computing, autonomous driving technology is developing rapidly and has become the latest development direction in intelligent transportation systems and intelligent vehicle engineering. Autonomous vehicles have a positive impact on transportation systems, mainly in the following ways: 1) energy conservation and emission reduction; 2) alleviating traffic congestion; 3) improving road capacity; and 4) freeing drivers' hands and reducing driver fatigue. However, the safety and system reliability of autonomous vehicles are not yet fully guaranteed. Therefore, to meet safety requirements, comprehensive evaluation of autonomous vehicles is necessary.

[0003] Simulation plays a crucial role in the evaluation of autonomous driving. Currently, there are two main types of simulation platforms on the market: Automotive Engineer Developed Platform (AEDP) and Traffic Engineer Developed Platform (TEDP). AEDP is equipped with realistic vehicle controllers and vehicle dynamics models, and can simulate real autonomous driving functions, but its simulation speed is slow and it cannot simulate new hybrid traffic flow scenarios consisting of autonomous driving and human driving. TEDP can simulate new hybrid traffic flow scenarios and supports real-time simulation, but it cannot simulate realistic vehicle controllers and dynamics models.

[0004] On the other hand, new hybrid traffic flows consisting of autonomous vehicles and human-driven vehicles will persist for at least the next 10 years. This new hybrid traffic will inevitably lead to frequent interactive games between autonomous vehicles and human-driven vehicles. Such a highly competitive traffic environment will result in a loss of driving efficiency for autonomous vehicles or an increase in safety hazards. However, existing driver models often make too many assumptions and simplifications, making it difficult to accurately characterize interactive game behavior. Therefore, to meet the testing needs of autonomous driving in new hybrid traffic environments, there is an urgent need to construct microscopic driving models that consider vehicle interactions, thereby simulating driving environments with highly competitive interactions. Summary of the Invention

[0005] The purpose of this invention is to provide an autonomous driving test system in a highly interactive game environment, which can simulate the interactive game behavior between autonomous vehicles and human-driven vehicles, thereby simulating a highly interactive traffic environment in the real world. At the same time, it can improve the simulation speed, meet the requirements of real-time simulation, and not only support the evaluation of autonomous driving functions, but also assess the impact of autonomous vehicles on the traffic system.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] An autonomous driving testing system in a highly competitive interactive environment includes:

[0008] Traffic simulator: It consists of six sub-units: road network construction, traffic generation, traffic control, driver model, traffic assessment, and information visualization, and is used to simulate traffic systems;

[0009] Interactive Game-Theoretic Decision Maker: The interactive game-theoretic decision maker is designed to meet the needs of interactive vehicle decision simulation. It takes the road traffic information output by the traffic simulator as input, embeds a game-theoretic decision algorithm based on logical rules, considers the scenarios of vehicles running on highways and urban roads, divides the vehicle operation mode into five categories: cruising, lane changing, returning to the lane, traffic lights, and conflict zones, constructs the decision mechanism for each operation mode, determines the operation mode of the next simulation step, and outputs the corresponding vehicle status.

[0010] An interactive game-theoretic controller is used to output control quantities for controlling the lateral and longitudinal behavior of an autonomous vehicle based on the vehicle state output by an interactive game-theoretic decision-maker, thereby realizing the game-theoretic struggle between the autonomous vehicle and the human-driven vehicle. The interactive game-theoretic controller includes an optimization model building subunit, a reaction function construction subunit, and an interactive game-theoretic algorithm design subunit.

[0011] The vehicle dynamics model is used to simulate the real response of the vehicle's mechanical system to the output of the interactive game-theoretic controller. The inputs are throttle opening, braking pressure and front wheel deflection angle control quantities, the output is the real state of the autonomous vehicle, and the feedback is given to the information visualization subunit.

[0012] In the traffic simulator unit, the road network construction subunit generates a road network; the traffic generation subunit generates a mixed traffic flow including human-driven vehicles and autonomous vehicles; the traffic control subunit formulates traffic management plans, including traffic light control plans, road control plans, and vehicle speed control plans; the driver model subunit controls the lateral and longitudinal behavior of human-driven vehicles and supports the embedding of custom models; the traffic evaluation subunit evaluates the macroscopic performance indicators of the traffic system, including average speed, density, delay, and queue length; and the information visualization subunit displays the simulation process through 2D or 3D animation and the simulation results through charts.

[0013] The decision-making mechanism in cruise mode is as follows:

[0014] If the current simulation step is in cruise mode, then proceed to the forced lane change judgment: if the conditions for a forced lane change are met simultaneously, the simulation step is in lane change mode, the simulation step is in lane change mode; otherwise...

[0015] Entering Free Lane Change Decision: If the following conditions are met simultaneously, the free lane change rule is satisfied, and the lane change safety distance rule is satisfied, then the operation mode for the next simulation step is lane change; otherwise...

[0016] Traffic light mode entry determination: If a vehicle enters the traffic light area but does not meet the intersection entry rules, the next simulation step's operating mode is traffic light; otherwise...

[0017] Conflict Zone Entry Mode Judgment: If the following conditions are met simultaneously: entering the conflict zone, not having priority, the presence of conflicting vehicles, and meeting the conflict event occurrence rules, then the operation mode of the next simulation step is the conflict zone; otherwise, the operation mode of the next simulation step is cruise.

[0018] The lane-changing safety distance rule stipulates that the headway between the vehicle and the vehicles in front and behind in the target lane must meet the following conditions:

[0019]

[0020] H TFV H TRV These represent the distance between the front of this vehicle and the vehicles in front and behind in the target lane, v ego v TFV v TRV These represent the speeds of the vehicle itself, the vehicle in front in the target lane, and the vehicle behind in the target lane, respectively. safe (v ego ,v TFV ), D safe (v TRV ,v egoThese represent the safe front-end distances between your vehicle and the vehicles in front and behind you in the target lane. The formula for calculating the safe distance is as follows:

[0021]

[0022] Where v0 and v1 are the speeds of the current vehicle and the target vehicle, respectively, t1 is the sum of the reaction time and braking delay time, s0 is the safe distance, and b max This is the maximum braking deceleration;

[0023] The free lane-changing rules include the following speed and space conditions:

[0024] v TFV -v ego ≥v diff

[0025] H TFV -H FV ≥D diff

[0026] Among them, v diff The speed threshold for adjacent lanes during free lane changing is defined as follows: when the speed difference between the vehicle in the target lane and the vehicle's speed exceeds the speed threshold, the adjacent lane speed condition for free lane changing is met. TFV H FV These are the distances between the front of this vehicle and the vehicle in front in the target lane, and the distance between the front of this vehicle and the vehicle in front in the current lane, respectively. diff The threshold for the distance between adjacent lanes during free lane changing is satisfied when the difference between the distance between the front of the vehicle and the vehicle in front of the target lane and the vehicle in front of the vehicle in the current lane exceeds the distance threshold.

[0027] The intersection entry rules are used to determine whether an autonomous vehicle can enter the signalized intersection. An intersection entry rule is satisfied when one of the following conditions is met: i) the traffic light is green; ii) the traffic light is yellow, and the following conditions are also met:

[0028] t2×v ego >D signal Or t yellow ×v ego >D signal

[0029] Where t2, t yellow These represent the control delay and the remaining time of the yellow light, respectively. signal This refers to the distance between the front of the vehicle and the traffic light.

[0030] The rules for the occurrence of conflict events are as follows:

[0031] t conflict -t ego ≤t diff

[0032] Among them, t conflict , t ego The times t represent the arrival times of the conflicting vehicles and the vehicle itself at the conflict zone. diff The time threshold is defined as the time difference between the arrival of the conflicting vehicle and the vehicle itself at the conflict zone being less than the time threshold, thus satisfying the conditions for a conflict to occur.

[0033] The decision-making mechanism under lane-changing mode is as follows:

[0034] If the current simulation step is in lane-changing mode, determine whether the vehicle is in the target lane and meets the lane-changing completion rules. If so, the next simulation step's operating mode is cruise; otherwise,

[0035] Determine whether the lane-changing safety distance rule is met. If yes, the next simulation step will be a lane-changing operation mode; otherwise, the next simulation step will be a return-to-lane operation mode.

[0036] The lane change completion rule is as follows:

[0037] |y ego -y tar |≤y diff1

[0038] Among them, y ego y tar Let y be the lateral coordinates of the vehicle and the target lane location in the road coordinate system. diff1 The allowable lateral deviation when a lane change is completed is defined as follows: the lane change is completed when the absolute value of the difference between the lateral coordinates of the vehicle and the target position is less than the allowable lateral deviation.

[0039] The decision-making mechanism in the return lane mode is as follows:

[0040] If the current simulation step is in lane return mode, determine whether the vehicle is in the original lane and meets the lane return completion rules. If so, the next simulation step will be in cruise mode; otherwise, the next simulation step will be in lane return mode.

[0041] The rules for completing the return lane are as follows:

[0042] |y ego -y ori |≤y diff2

[0043] Among them, y ego y ori Let y be the lateral coordinates of the vehicle and the starting position of the lane change in the road coordinate system. diff2 The allowable lateral deviation when returning to the lane is completed is defined as follows: when the absolute value of the difference between the lateral coordinates of the vehicle and the starting position of the lane change is less than the allowable lateral deviation, the vehicle returns to the lane.

[0044] The decision-making mechanism under traffic light mode is as follows:

[0045] If the current simulation step is in traffic light mode, check the traffic light status. If the traffic light is not green, the next simulation step will be in traffic light mode. If the traffic light is green, enter the conflict zone mode. If the following conditions are met simultaneously: entering the conflict zone, not having priority, having conflicting vehicles, and meeting the conflict event occurrence rules, the next simulation step will be in conflict zone mode. Otherwise, the next simulation step will be in cruise mode.

[0046] The decision-making mechanism under the conflict zone model is as follows:

[0047] If the current simulation step is in conflict zone mode, and simultaneously meets the conditions of entering the conflict zone, not having priority, having conflicting vehicles, and meeting the conflict event occurrence rules, then the next simulation step will operate in conflict zone mode; otherwise, the next simulation step will operate in cruise mode.

[0048] The optimization model building subunit is used to establish an optimization control model that considers the interactive game between the two vehicles. Considering that both the driver and target vehicles have a cost function to be optimized, and integrating the kinematic models of the driver and target vehicles into a single system dynamic equation, the form of the optimization control model is as follows:

[0049]

[0050]

[0051] stx k+1 =A k x k +B k u 1,k +C k u 2,k

[0052] Where J1 represents the cost function of the main vehicle and J2 represents the cost function of the target vehicle; The weighting parameters for the main vehicle; The weight parameters for the target vehicle; These are system state variables, including the longitudinal distance, lateral distance, vehicle speed, and heading angle between the host vehicle and the target vehicle. These represent the desired driving states for the main vehicle and the target vehicle, respectively. The control inputs are for the master vehicle and the target vehicle, respectively, including the desired acceleration and the desired front wheel deflection angle.

[0053] The reaction function construction subunit is used to establish the target vehicle driver's reaction function, which is the mapping relationship between the system state and the master vehicle control input and the target vehicle control input at each time step. The mathematical expression of the reaction function is as follows:

[0054] u 2,k =J k+1 x k KG k+1 u 1,k +F k+1 +E k+1

[0055] in, All are given by the following formulas:

[0056] J k+1 =H k+1 P k+1 A k

[0057] G k+1 =H k+1 P k+1 C k

[0058] F k+1 =-H k+1 P k+1 x 2,des

[0059] E k+1 =H k+1 O k+1

[0060]

[0061] Among them, A k For the system matrix, B k C k For the control matrix, P k With O k Determined by the following recursive formula and initial conditions:

[0062]

[0063]

[0064] P N =Q2,O N =0

[0065] Where I is the identity matrix.

[0066] The interactive game-theoretic algorithm design subunit is used to implement the lateral and longitudinal control of autonomous vehicles that considers interactive game theory. The interactive game-theoretic algorithm uses the driver's reaction function as a constraint, the driver's multi-objective optimization control model as a cost function, and uses quadratic programming to obtain the optimal control strategy for autonomous driving.

[0067] Specifically, the interactive game theory algorithm rewrites the optimal control model J1 of the master vehicle into a sequential form, and substitutes the reaction function of the target vehicle into the optimal control model J1 of the master vehicle to form a quadratic programming form with linear constraints. The Lagrange multiplier method is then used to solve the model, and the solution is expressed as follows:

[0068]

[0069] Among them, [X,U 1 ,λ] T The optimal solution to the control problem includes the optimal trajectory X of the main vehicle and the desired control input U. 1 And the Lagrange multiplier λ, and All are coefficient matrices.

[0070] Compared with the prior art, the present invention has the following beneficial effects:

[0071] (1) Compared with existing simulation test platforms, this invention integrates the advantages of car simulation platforms and traffic simulation platforms and upgrades them to form a universal autonomous driving test system. This invention not only embeds a highly realistic vehicle dynamics model as well as an autonomous driving interactive game-like decision-maker and controller, which can simulate autonomous driving functions, but also supports a new type of hybrid traffic simulation consisting of large-scale autonomous driving and human driving, meeting the needs of real-time simulation.

[0072] (2) This invention describes the game-like interaction mechanism between autonomous driving and human-driven vehicle behavior decisions, constructs a vehicle-to-vehicle interaction behavior model, and simulates a real-world traffic environment with strong game-like interaction, providing a realistic environment for the simulation test of autonomous vehicles and improving the credibility of the simulation test.

[0073] (3) Based on optimization control and game theory, this invention constructs a micro-driving behavior game interaction model between autonomous driving and human-driven vehicles, characterizes the game interaction influence mechanism between the two, and realizes the game struggle between autonomous driving vehicles and human-driven vehicles.

[0074] (4) This invention proposes an interactive game-theoretic decision-maker for autonomous driving. This decision-maker is designed to meet the needs of vehicle interactive decision simulation. It embeds a game-theoretic decision-making algorithm based on logical synthesis, considers the scenarios of operation on highways and urban roads, and constructs vehicle interactive decision-making mechanisms in multiple scenarios such as cruising, lane changing, traffic lights and conflict zones, thereby more clearly depicting vehicle decision-making behavior.

[0075] (5) This invention proposes an interactive game-theoretic controller for autonomous driving. This controller overcomes the shortcomings of current autonomous driving technologies that only consider the vehicle's state and benefits. It predicts the reactions of surrounding human-driven vehicles to autonomous driving behavior through interactive game theory, and then makes the optimal driving control for the vehicle based on this prediction.

[0076] (6) In the autonomous driving test layer, the present invention meets the testing needs of both vehicle engineers and traffic engineers: For vehicle engineers, the present invention is compatible with a variety of autonomous driving algorithms and vehicle dynamics models, and can support autonomous driving function testing and expected functional safety testing; For traffic engineers, the present invention can flexibly modify parameters such as traffic environment and traffic flow, and can evaluate the impact of autonomous vehicles on the traffic system. Attached Figure Description

[0077] Figure 1 This is a schematic diagram of the system structure of the present invention;

[0078] Figure 2 A schematic diagram illustrating the decision-making mechanism of an autonomous vehicle in cruise mode;

[0079] Figure 3 A schematic diagram illustrating the decision-making mechanism of an autonomous vehicle in lane-changing mode;

[0080] Figure 4 A schematic diagram illustrating the decision-making mechanism for an autonomous vehicle in lane-return mode;

[0081] Figure 5 A schematic diagram of the decision-making mechanism structure for an autonomous vehicle in traffic light mode;

[0082] Figure 6 A schematic diagram of the decision-making mechanism structure for autonomous vehicles in conflict zone mode;

[0083] Figure 7 This is a schematic diagram of the vehicle dynamics model. Detailed Implementation

[0084] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0085] This embodiment provides an autonomous driving testing system in a highly competitive interactive environment, such as... Figure 1 As shown, it includes:

[0086] (1) Traffic simulator

[0087] Traffic simulators are used to simulate traffic systems and consist of six sub-units: road network construction, traffic generation, traffic control, driver modeling, traffic assessment, and information visualization.

[0088] The road network construction subunit generates a road network; the traffic generation subunit generates a mixed traffic flow including both human-driven and autonomous vehicles; the traffic control subunit formulates traffic control schemes, including but not limited to traffic light control schemes, road control schemes, and vehicle speed control schemes; the driver model subunit controls the lateral and longitudinal behavior of human-driven vehicles and supports the embedding of custom models; the traffic assessment subunit evaluates the macroscopic performance indicators of the traffic system, including but not limited to average vehicle speed, density, delay, and queue length; and the information visualization subunit displays the simulation process through 2D or 3D animations and the simulation results through charts.

[0089] (2) Interactive game-theoretic decision-making system

[0090] The interactive game-theoretic decision-maker addresses the needs of interactive vehicle decision-making simulation. It takes road traffic information output by a traffic simulator as input, embeds a game-theoretic decision-making algorithm based on logical rules, considers the scenarios of vehicles running on highways and urban roads, divides vehicle operation modes into five categories: cruising, lane changing, returning to the lane, traffic lights, and conflict zones, constructs a decision-making mechanism for each operation mode, determines the operation mode of the next simulation step, and outputs the corresponding vehicle status.

[0091] ① Cruise mode

[0092] The decision-making mechanism in cruise mode determines the operating mode of the autonomous vehicle currently in cruise mode in the next simulation step, such as... Figure 2 As shown, specifically:

[0093] If the current simulation step is in cruise mode, then proceed to the forced lane change judgment: if the conditions for a forced lane change are met simultaneously, the simulation step is in lane change mode, the simulation step is in lane change mode; otherwise...

[0094] Entering Free Lane Change Decision: If the following conditions are met simultaneously, the free lane change rule is satisfied, and the lane change safety distance rule is satisfied, then the operation mode for the next simulation step is lane change; otherwise...

[0095] Traffic light mode entry determination: If a vehicle enters the traffic light area but does not meet the intersection entry rules, the next simulation step's operating mode is traffic light; otherwise...

[0096] Conflict Zone Entry Mode Judgment: If the following conditions are met simultaneously: entering a conflict zone, not having priority, the presence of conflicting vehicles, and meeting the conflict event occurrence rules, then the operation mode of the next simulation step is conflict zone; otherwise, the operation mode of the next simulation step is cruise.

[0097] ② Lane changing mode

[0098] The decision-making mechanism in lane-changing mode determines the operating mode of an autonomous vehicle currently in lane-changing mode in the next simulation step, such as... Figure 3 As shown, specifically:

[0099] If the current simulation step is in lane-changing mode, determine whether the vehicle is in the target lane and meets the lane-changing completion rules. If so, the next simulation step's operating mode is cruise; otherwise,

[0100] Determine whether the lane-changing safety distance rule is met. If so, the next simulation step's operating mode is lane changing; otherwise, the next simulation step's operating mode is returning to the lane.

[0101] ③ Return Lane Mode

[0102] The decision-making mechanism in the lane-return mode determines the operating mode of the autonomous vehicle currently in lane-return mode in the next simulation step, such as... Figure 4 As shown, specifically:

[0103] If the current simulation step is in lane return mode, determine whether the vehicle is in the original lane and meets the lane return completion rules. If so, the next simulation step's operating mode is cruise; otherwise, the next simulation step's operating mode is lane return.

[0104] ④ Traffic light mode

[0105] The decision-making mechanism in traffic light mode determines the operating mode of an autonomous vehicle currently in traffic light mode in the next simulation step, such as... Figure 5 As shown, specifically:

[0106] If the current simulation step is in traffic light mode, check the traffic light status. If the traffic light is not green, the next simulation step will be in traffic light mode. If the traffic light is green, enter the conflict zone mode. If the following conditions are met simultaneously: entering the conflict zone, not having priority, having conflicting vehicles, and meeting the conflict event occurrence rules, the next simulation step will be in conflict zone mode. Otherwise, the next simulation step will be in cruise mode.

[0107] ⑤ Conflict Zone Model

[0108] The decision-making mechanism in the conflict zone mode determines the operating mode of the autonomous vehicle currently in the conflict zone mode in the next simulation step, such as... Figure 6 As shown, specifically:

[0109] If the current simulation step is in conflict zone mode, and simultaneously meets the conditions of entering the conflict zone, not having priority, having conflicting vehicles, and meeting the conflict event occurrence rules, then the next simulation step will operate in conflict zone mode; otherwise, the next simulation step will operate in cruise mode.

[0110] The specific rules involved in the above decision-making mechanism are as follows:

[0111] a) Lane changing safe distance rules

[0112] Lane-changing safety distance rules are used to determine whether lane-changing distances are safe for autonomous driving. When autonomous driving safely changes lanes, the headway between the vehicle and the vehicles in front and behind in the target lane must meet the following conditions:

[0113]

[0114] Among them, H TFV H TRV These represent the distance between the front of this vehicle and the vehicles in front and behind in the target lane, v ego v TFV v TRV These represent the speeds of the vehicle itself, the vehicle in front in the target lane, and the vehicle behind in the target lane, respectively. safe (v ego ,v TFV ), D safe (v TRV ,v ego These represent the safe front-end distances between your vehicle and the vehicles in front and behind you in the target lane. The formula for calculating the safe distance is as follows:

[0115]

[0116] Where v0 and v1 are the speeds of the current vehicle and the target vehicle, respectively, t1 is the sum of the reaction time and braking delay time, s0 is the safe distance, and b max This is the maximum braking deceleration.

[0117] b) Free lane changing rules

[0118] The free lane-changing rule determines whether an autonomous vehicle can perform a free lane change. Free lane changing requires the following speed and space conditions to be met:

[0119] v TFe -v ego ≥v diff

[0120] HTFV -H FV ≥D diff

[0121] Among them, v diff The speed threshold for adjacent lanes during free lane changing is defined as follows: when the speed difference between the vehicle in the target lane and the vehicle's speed exceeds the speed threshold, the adjacent lane speed condition for free lane changing is met. TFV H FV These are the distances between the front of this vehicle and the vehicle in front in the target lane, and the distance between the front of this vehicle and the vehicle in front in the current lane, respectively. diff The threshold for the distance between adjacent lanes during free lane changing is satisfied when the difference between the distance between the front of the vehicle and the vehicle in front of the target lane and the vehicle in front of the vehicle in the current lane exceeds the distance threshold.

[0122] c) Rules for entering intersections

[0123] Entry rules for intersections determine whether autonomous vehicles can enter a signalized intersection. Autonomous vehicles can cross the stop line and enter an intersection if one of the following two conditions is met:

[0124] i) The traffic light is green;

[0125] ii) The traffic light is yellow, and the following conditions are met:

[0126] t2×v ego >D signal Or t yellow ×v ego >D signal

[0127] Where t2, t yellow These represent the control delay and the remaining time of the yellow light, respectively. signal This refers to the distance between the front of the vehicle and the traffic light.

[0128] d) Rules for the occurrence of conflict events

[0129] Conflict event rules are used to determine whether a conflict event exists for autonomous vehicles. A conflict event must meet the following conditions to occur:

[0130] t conflict -t ego ≤t diff

[0131] Among them, t conflict , t ego The times t represent the arrival times of the conflicting vehicles and the vehicle itself at the conflict zone. diff The time threshold is defined as the time difference between the arrival of the conflicting vehicle and the vehicle itself at the conflict zone being less than the time threshold, thus satisfying the conditions for a conflict to occur.

[0132] e) Lane change completion rules

[0133] Lane change completion rules are used to determine whether an autonomous vehicle has completed a lane change. A lane change must meet the following conditions to be considered complete:

[0134] |y ego -y tar |≤y diff1

[0135] Among them, y ego y tar Let y be the lateral coordinates of the vehicle and the target lane location in the road coordinate system. diff1 The allowable lateral deviation when a lane change is completed is defined as follows: the lane change is completed when the absolute value of the difference between the lateral coordinates of the vehicle and the target position is less than the allowable lateral deviation.

[0136] f) Return to lane completion rules

[0137] The lane return completion rule determines whether an autonomous vehicle returns to its original lane at the lane change starting point. The vehicle's return to its original lane must meet the following conditions:

[0138] |y ego -y ori |≤y diff2

[0139] Among them, y ego y ori Let y be the lateral coordinates of the vehicle and the starting position of the lane change in the road coordinate system. diff2 The allowable lateral deviation when returning to the lane is completed is defined as follows: when the absolute value of the difference between the lateral coordinates of the vehicle and the starting position of the lane change is less than the allowable lateral deviation, the vehicle returns to the lane.

[0140] (3) Interactive game-theoretic controller

[0141] The interactive game-theoretic controller is used to output control quantities for the lateral and longitudinal behavior of autonomous vehicles based on the vehicle state output by the interactive game-theoretic decision-maker. This enables autonomous vehicles to engage in a game-theoretic struggle with human-driven vehicles, improving their maneuverability while ensuring traffic safety.

[0142] In this embodiment, the interactive game-theoretic controller includes an optimization model building subunit, a reaction function construction subunit, and an interactive game-theoretic algorithm design subunit.

[0143] (31) Optimize the model building sub-unit

[0144] The optimization model building subunit is used to establish an optimization control model that considers the interactive game between the two vehicles. Considering that both the master vehicle (autonomous vehicle) and the target vehicle (human-driven vehicle) have a cost function to be optimized, and integrating the kinematic models of the master and target vehicles into a single system dynamic equation, the form of the optimization control model is as follows:

[0145]

[0146]

[0147] stx k+1 =A k x k +B k u 1,k +C k u 2,k

[0148] Where J1 represents the cost function of the main vehicle and J2 represents the cost function of the target vehicle; The weighting parameters for the main vehicle; The weight parameters for the target vehicle; These are system state variables, including the longitudinal distance, lateral distance, vehicle speed, and heading angle between the host vehicle and the target vehicle. These represent the desired driving states for the main vehicle and the target vehicle, respectively. The control inputs are for the master vehicle and the target vehicle, respectively, including the desired acceleration and the desired front wheel deflection angle.

[0149] (32) Reaction function constructs subunits

[0150] The reaction function construction sub-unit is used to establish the target vehicle driver's reaction function, which is the mapping relationship between the system state and the master vehicle control input and the target vehicle control input at each time step. The mathematical expression of the reaction function is as follows:

[0151] u 2,k =J k+1 x k +G k+1 u 1,k +F k+1 +E k+1

[0152] in, All are given by the following formulas:

[0153] J k+1 =H k+1 P k+1 A k

[0154] Gk+1 =H k+1 P k+1 C k

[0155] F k+1 =-H k+1 P k+1 x 2,des

[0156] E k+1 =H k+1 O k+1

[0157]

[0158] The above formulas not only depend on the system matrix A k and control matrix B k C k It also depends on matrix P k With O k P k With O k Determined by the following recursive formula and initial conditions:

[0159]

[0160]

[0161] P N =Q2,O N =0

[0162] Where I is the identity matrix.

[0163] (33) Interactive game-theoretic algorithm design subunit

[0164] The interactive game-theoretic algorithm design subunit is used to implement the lateral and longitudinal control of autonomous vehicles that considers interactive game theory. The interactive game-theoretic algorithm uses the driver's reaction function as a constraint, the driver's multi-objective optimization control model as a cost function, and the quadratic programming method to obtain the optimal control strategy for autonomous driving.

[0165] Specifically, the interactive game theory algorithm rewrites the optimal control model J1 of the master vehicle into a sequential form, and substitutes the reaction function of the target vehicle into the optimal control model J1 of the master vehicle to form a quadratic programming form with linear constraints. The Lagrange multiplier method is then used to solve the model, and the solution is expressed as follows:

[0166]

[0167] Among them, [X,U 1 ,λ] TThe optimal solution to the control problem includes the optimal trajectory X of the main vehicle and the desired control input U. 1 And the Lagrange multiplier λ, and All are coefficient matrices.

[0168] (4) Vehicle dynamics model

[0169] The vehicle dynamics model is used to simulate the real response of the vehicle's mechanical system to the output of the interactive game-like controller. The inputs are throttle opening, braking pressure and front wheel deflection angle control quantities. The output is the real state of the autonomous vehicle, including but not limited to speed, position and heading angle, and is fed back to the information visualization subunit of the traffic simulator.

[0170] like Figure 7 As shown, the vehicle dynamics model consists of five sub-units: transmission, engine, drivetrain, chassis, and steering system.

[0171] (41) Transmission subunit

[0172] The transmission subunit calculates the gear based on speed and throttle opening. The gear is then determined by the shift gauge within this subunit.

[0173] (42) Engine Subunit

[0174] The engine subunit calculates engine torque, engine angular velocity, and engine power. Engine torque is calculated using a first-order response function.

[0175]

[0176] Where k4 is the static gain, θ is the damping coefficient, and ω n2 It is a natural frequency. It's a time delay.

[0177] The engine angular velocity is calculated by solving the engine power equation:

[0178]

[0179] Among them, P e It is engine power, ω e It is the engine angular velocity, P M It is the engine's maximum power, ω M It is the engine's maximum angular velocity, P e It is calculated from the following formula:

[0180] P e =T e ω e

[0181] Where T eIt refers to engine torque.

[0182] (43) Transmission device subunit

[0183] The transmission subunit takes engine speed and gear as input and outputs actual engine torque.

[0184]

[0185] Among them, C c It is the slip coefficient, ρ o It is the density of gasoline, ω p D is the angular velocity of the oil pump, and D is the diameter of the clutch pressure plate.

[0186] (44) Base plate unit

[0187] The chassis unit outputs the vehicle speed, calculated using the following formula:

[0188]

[0189] Among them, R w n is the effective radius of the tire. g It is the gearbox ratio, n d It is the differential gear ratio.

[0190] (45) Steering System Subunit

[0191] The steering system subunit takes the front wheel deflection angle as input and outputs the vehicle heading angle. The formula for the rate of change of the heading angle is as follows:

[0192]

[0193] in, β is the rate of change of heading angle, β is the vehicle's center of gravity deflection angle (the angle between the velocity direction and the longitudinal plane of symmetry of the vehicle body), and δ is the yaw rate. f It is the front wheel deflection angle, w is the vehicle wheelbase, and k is the road curvature.

[0194] In summary, the autonomous driving testing system proposed in this embodiment under a strong game-theoretic interactive environment has at least the following advantages:

[0195] 1. It can simulate a real-world traffic environment with strong game-theoretic interactions, providing a realistic environment for the simulation testing of autonomous vehicles and improving the credibility of simulation tests;

[0196] 2. It adopts a modular design structure, is compatible with autonomous driving decision-makers, controllers, and dynamic models, and supports testing and verification of various autonomous driving functions;

[0197] 3. It can generate novel hybrid traffic flows consisting of large-scale autonomous driving and human driving, supporting the assessment of the impact of autonomous driving on the traffic system;

[0198] 4. Based on optimization control and game theory, a micro-level driving behavior game interaction model between autonomous driving and human-driven vehicles was constructed, which describes the game interaction influence mechanism between the two and realizes the game struggle between autonomous driving vehicles and human-driven vehicles.

[0199] 5. The simulation speed is fast, and it can simulate 300 autonomous vehicles plus an unlimited number of human-driven vehicles in real time.

[0200] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. An autonomous driving testing system under a strong game-theoretic interactive environment, characterized in that, include: Traffic simulator: It consists of six sub-units: road network construction, traffic generation, traffic control, driver model, traffic assessment, and information visualization, and is used to simulate traffic systems; Interactive Game-Theoretic Decision Maker: The interactive game-theoretic decision maker is designed to meet the needs of interactive vehicle decision simulation. It takes the road traffic information output by the traffic simulator as input, embeds a game-theoretic decision algorithm based on logical rules, considers the scenarios of vehicles running on highways and urban roads, divides the vehicle operation mode into five categories: cruising, lane changing, returning to the lane, traffic lights, and conflict zones, constructs the decision mechanism for each operation mode, determines the operation mode of the next simulation step, and outputs the corresponding vehicle status. An interactive game-theoretic controller is used to output control quantities for controlling the lateral and longitudinal behavior of an autonomous vehicle based on the vehicle state output by an interactive game-theoretic decision-maker, thereby realizing the game-theoretic struggle between the autonomous vehicle and the human-driven vehicle. The interactive game-theoretic controller includes an optimization model building subunit, a reaction function construction subunit, and an interactive game-theoretic algorithm design subunit. The vehicle dynamics model is used to simulate the real response of the vehicle's mechanical system to the output of the interactive game-like controller. The inputs are throttle opening, braking pressure and front wheel deflection angle control quantities, the output is the real state of the autonomous vehicle, and the feedback is given to the information visualization subunit. The optimization model building subunit is used to establish an optimization control model that considers the interactive game between the two vehicles. Considering that both the main vehicle and the target vehicle have a cost function to be optimized, and integrating the kinematic models of the main vehicle and the target vehicle into a single system dynamic equation, the form of the optimization control model is as follows: in, This represents the cost function of the main vehicle. Represented as the cost function of the target vehicle; , The weighting parameters for the main vehicle; , The weight parameters for the target vehicle; These are system state variables, including the longitudinal distance, lateral distance, vehicle speed, and heading angle between the host vehicle and the target vehicle. , These represent the desired driving states for the main vehicle and the target vehicle, respectively. , The control inputs for the master vehicle and the target vehicle are respectively, including the desired acceleration and the desired front wheel deflection angle; The reaction function construction subunit is used to establish the target vehicle driver's reaction function, which is the mapping relationship between the system state and the master vehicle control input and the target vehicle control input at each time step. The mathematical expression of the reaction function is as follows: in, , , , All are given by the following formulas: in, For the system matrix, , For the control matrix, and Determined by the following recursive formula and initial conditions: in, It is the identity matrix; The interactive game-theoretic algorithm design subunit is used to implement the lateral and longitudinal control of autonomous vehicles that considers interactive game theory. The interactive game-theoretic algorithm uses the driver's reaction function as a constraint, the driver's multi-objective optimization control model as a cost function, and uses quadratic programming to obtain the optimal control strategy for autonomous driving. Specifically, the interactive game theory algorithm will optimize the control model of the master vehicle. Rewrite it in sequential form and substitute the target vehicle's reaction function into the master vehicle's optimal control model. This forms a quadratic programming problem with linear constraints, and the Lagrange multiplier method is used to solve the model. The solution is expressed as follows: in, To optimize the optimal solution to the control problem, the optimal trajectory of the main vehicle is included. and expected control input and Lagrange multipliers , , , and All are coefficient matrices.

2. The autonomous driving test system under a strong game-theoretic interactive environment according to claim 1, characterized in that, In the traffic simulator, the road network construction subunit generates a road network; the traffic generation subunit generates a mixed traffic flow including human-driven vehicles and autonomous vehicles; the traffic control subunit formulates traffic management plans, including traffic light control plans, road management plans, and vehicle speed control plans; the driver model subunit controls the lateral and longitudinal behavior of human-driven vehicles and supports the embedding of custom models; the traffic evaluation subunit evaluates the macroscopic performance indicators of the traffic system, including average speed, density, delay, and queue length; and the information visualization subunit displays the simulation process through 2D or 3D animation and the simulation results through charts.

3. The autonomous driving test system under a strong game-theoretic interactive environment according to claim 1, characterized in that, The decision-making mechanism in cruise mode is as follows: If the current simulation step is in cruise mode, then proceed to the forced lane change judgment: if the conditions for a forced lane change are met simultaneously, the simulation step is in lane change mode, the simulation step is in lane change mode; otherwise, Entering the free lane change judgment: If the following state, the free lane change rule, and the lane change safety distance rule are all satisfied simultaneously, then the operation mode of the next simulation step is lane change; otherwise, Traffic light mode entry determination: If a vehicle enters the traffic light area but does not meet the intersection entry rules, the next simulation step's operating mode is traffic light; otherwise, Conflict Zone Entry Mode Judgment: If the following conditions are met simultaneously: entering the conflict zone, not having priority, the presence of conflicting vehicles, and meeting the conflict event occurrence rules, then the operation mode of the next simulation step is the conflict zone; otherwise, the operation mode of the next simulation step is cruise. The lane-changing safety distance rule stipulates that the headway between the vehicle and the vehicles in front and behind in the target lane must meet the following conditions: , These are the distances between the front and rear of the vehicle and the vehicles in front and behind in the target lane, respectively. , , These are the speeds of the vehicle in the target lane, the vehicle in front of the vehicle in the target lane, and the vehicle behind the vehicle in the target lane, respectively. , These are the safe front-end distances between your vehicle and the vehicles in front and behind you in the target lane. The formula for calculating the safe distance is as follows: in, , These are the speeds of the current vehicle and the target vehicle, respectively. It is the sum of reaction time and braking delay time. For a safe distance, This is the maximum braking deceleration; The free lane-changing rules include the following speed and space conditions: in, The speed threshold for adjacent lanes during free lane changing is defined as follows: when the speed difference between the vehicle in the target lane and the current vehicle exceeds the speed threshold, the adjacent lane speed condition for free lane changing is met. , These are the distances between the front of this vehicle and the vehicle in front in the target lane, and the distances between the front of this vehicle and the vehicle in front in the current lane, respectively. The threshold for the distance between adjacent lanes during free lane changing is satisfied when the difference between the distance between the front of the vehicle and the vehicle in front of the target lane and the vehicle in front of the vehicle in the current lane exceeds the distance threshold. The intersection entry rules are used to determine whether an autonomous vehicle can enter the signalized intersection. An intersection entry rule is satisfied when one of the following conditions is met: i) the traffic light is green; ii) the traffic light is yellow, and the following conditions are also met: in, , These are the control delay and the remaining time of the yellow light, respectively. This refers to the distance between the front of the vehicle and the traffic light. The rules for the occurrence of conflict events are as follows: in, , These are the arrival times of the conflicting vehicle and the vehicle itself at the conflict zone. The time threshold is defined as the time difference between the arrival of the conflicting vehicle and the vehicle itself at the conflict zone being less than the time threshold, thus satisfying the conditions for a conflict to occur.

4. The autonomous driving test system under a strong game-theoretic interactive environment according to claim 3, characterized in that, The decision-making mechanism under lane-changing mode is as follows: If the current simulation step is in lane-changing mode, determine whether the vehicle is in the target lane and meets the lane-changing completion rules. If so, the next simulation step's operating mode is cruise; otherwise, Determine whether the lane-changing safety distance rule is met. If yes, the next simulation step will be a lane-changing operation mode; otherwise, the next simulation step will be a return-to-lane operation mode. The lane change completion rule is as follows: in, , Let be the lateral coordinates of the vehicle and the target lane change location in the road coordinate system. The allowable lateral deviation when a lane change is completed is defined as follows: the lane change is completed when the absolute value of the difference between the lateral coordinates of the vehicle and the target position is less than the allowable lateral deviation.

5. The autonomous driving test system under a strong game-theoretic interactive environment according to claim 1, characterized in that, The decision-making mechanism in the return lane mode is as follows: If the current simulation step is in lane return mode, determine whether the vehicle is in the original lane and meets the lane return completion rules. If so, the next simulation step will be in cruise mode; otherwise, the next simulation step will be in lane return mode. The rules for completing the return lane are as follows: in, , Let be the lateral coordinates of the vehicle and the starting position of the lane change in the road coordinate system. The allowable lateral deviation when returning to the lane is completed is defined as follows: when the absolute value of the difference between the lateral coordinates of the vehicle and the starting position of the lane change is less than the allowable lateral deviation, the vehicle returns to the lane.

6. The autonomous driving test system under a strong game-theoretic interactive environment according to claim 3, characterized in that, The decision-making mechanism under traffic light mode is as follows: If the current simulation step is in traffic light mode, check the traffic light status. If the traffic light is not green, the next simulation step will be in traffic light mode. If the traffic light is green, enter the conflict zone mode. If the following conditions are met simultaneously: entering the conflict zone, not having priority, having conflicting vehicles, and meeting the conflict event occurrence rules, the next simulation step will be in conflict zone mode. Otherwise, the next simulation step will be in cruise mode.

7. The autonomous driving test system under a strong game-theoretic interactive environment according to claim 3, characterized in that, The decision-making mechanism under the conflict zone model is as follows: If the current simulation step is in conflict zone mode, and simultaneously meets the conditions of entering the conflict zone, not having priority, having conflicting vehicles, and meeting the conflict event occurrence rules, then the next simulation step will operate in conflict zone mode; otherwise, the next simulation step will operate in cruise mode.

Images