An intelligent vehicle personalized lane-changing decision system with weight self-evolution

Abstract

The present application relates to a kind of weight self-evolution intelligent car personalized lane change decision system, comprising: driving style quantification module, lane change decision module, lane change performance baseline evaluation module;Driving style quantification module gives driving style quantification result, output to lane change decision module, lane change decision module calculates the benefit function of SV and PV and the weight of the benefit function of PV and makes lane change decision judgment, lane change performance baseline evaluation module combines the decision result output by the lane change decision submodule of data-driven, assesses current more optimal decision result, and output evaluation result to weight self-learning submodule and baseline self-evaluator, weight self-learning submodule dynamically adjusts the weight of the benefit function of PV, and baseline self-evaluator determines whether the current weight self-evolution process is terminated according to evaluation result.The present application has autonomous learning ability, greatly improves the working condition adaptability and flexibility of lane change decision system, and can achieve good performance in various complex and changeable driving conditions.

Description

Technical Field

[0001] The present invention relates to a lane-changing behavior decision system for an autonomous driving vehicle, and in particular to a weighted self-evolving intelligent vehicle personalized lane-changing decision system. Background Art

[0002] As the level of automotive intelligence continues to improve, the production and sales of smart cars equipped with high-level autonomous driving systems are increasing. High-level autonomous driving systems should be able to perceive complex traffic situations and make accurate behavioral decisions in interaction with surrounding vehicles, thereby reducing collision risk while improving traffic efficiency. Lane changing is a core component of microscopic traffic flow and a typical vehicle-to-vehicle interaction. During this lane changing interaction, whether the vehicle can correctly understand the intentions of surrounding vehicles and make appropriate decisions is a major technical challenge in the current field of intelligent vehicle technology.

[0003] In order to characterize the personalized differences in driving behaviors of different drivers, a personalized lane-changing decision system for intelligent vehicles was established in combination with a driving style classification algorithm based on driving behavior. At present, personalized lane-changing decision systems for intelligent vehicles generally include two categories: data-driven lane-changing decisions and lane-changing decisions based on mechanism rules. Data-driven lane-changing decisions generally use learning methods to take features such as the kinematic parameters of the vehicle and surrounding vehicles as inputs to the learning algorithm, and use neural networks and other methods to fit the intrinsic mapping relationship between the input features and the vehicle's behavioral intentions; lane-changing decisions based on mechanism rules generally use manually set rules and common sense cognition to establish a mapping relationship between the driver's driving behavior and the driver's intentions. However, the data-driven and mechanism rule-based intelligent vehicle lane-changing behavior decision systems still have the following technical problems:

[0004] 1. Current driving style classification algorithms based on driving behavior are only capable of qualitative analysis and struggle with quantitative analysis. Currently constructed driving style classification algorithms categorize drivers' driving styles into only three categories: cautious, average, and aggressive. In reality, individual drivers' driving styles vary and are easily influenced by the traffic environment. Accurately assessing driving style is crucial for correctly judging the driver's driving intentions and making informed decisions. However, existing driving style classification algorithms struggle to provide quantitative metrics for driving style.

[0005] 2. Data-driven lane-changing decision systems establish a multidimensional mapping relationship between the kinematic variables of surrounding vehicles and the vehicle itself during driving and its lane-changing decisions. Essentially, this system learns the implicit interaction logic from massive amounts of data, thereby establishing an endogenous relationship between input parameters and output results. Because data-driven approaches are black-box algorithms with poor interpretability, it is difficult to explicitly express the interaction between system parameters and traffic flow during lane-changing decisions. This makes it difficult to adjust system parameters, leading to unstable system outputs and even safety risks.

[0006] 3. Mechanism-based lane-changing decision systems utilize observations of the driver's driving behavior to establish a meaningful functional mapping between vehicle kinematic information and vehicle driving intentions. Lane-changing decisions are made based on manually set decision rules, taking into account the vehicle's kinematic constraints and driving safety. However, the real-world traffic environment is complex and ever-changing, and manual rules struggle to cover all driving conditions. This results in the poor versatility of mechanism-based lane-changing decision systems. Furthermore, the parameters in mechanism-based lane-changing decision systems typically remain fixed, making it difficult for the system to acquire autonomous learning capabilities, further limiting the performance of mechanism-based lane-changing decision systems. Summary of the Invention

[0007] To address the above-mentioned technical problems, the present invention provides a weighted self-evolving personalized lane-changing decision system for intelligent vehicles, comprising: a driving style quantification module, a lane-changing decision module, and a lane-changing performance baseline assessment module; the driving style quantification module comprises a kinematic feature extraction submodule, a driving style assessment submodule, and a driving style quantification data temporary storage submodule; the lane-changing decision module comprises: a lane-changing benefit function submodule, a decision judgment submodule, a weighted self-learning submodule, and a driving style and dynamic weight storage submodule; the lane-changing performance baseline assessment module comprises a data-driven lane-changing decision submodule, a decision assessment submodule, and a baseline self-assessor.

[0008] Furthermore, the working process of the driving style quantification module is as follows: first, the kinematic information of the vehicle behind the target lane is transmitted to the kinematic feature extraction submodule, the kinematic feature extraction submodule performs data preprocessing and data feature extraction on the transmitted kinematic information, and then transmits the processed information to the driving style evaluation submodule. The trained Gaussian mixture model (GMM) in the driving style evaluation submodule receives the above information and, based on the obtained information, gives the quantified probability of each driving style of the vehicle behind the target lane, including: the probability of cautious type x d1 , the general probability x d2 , the probability of radical type x d3 , and the probability x d1 、x d2 、x d3Output to the driving style quantization data temporary storage submodule, which organizes the data into a vector X with one row and three columns d =[x d1 ,x d2 ,x d3 ], that is, driving style quantization data, and is stored in the data register; when new driving style quantization data is transmitted to the driving style quantization data temporary storage sub-module, the previous data stored in the data register will be overwritten by the new quantization data.

[0009] Specifically, the data feature extraction function of the kinematic feature extraction submodule is to obtain the relative speed, relative acceleration, and jerk information of the vehicle behind the target lane relative to the vehicle within time T, calculate the average relative speed, average relative acceleration, and average jerk, and transmit the calculated average relative speed, average relative acceleration, and average jerk information to the driving style assessment submodule.

[0010] Specifically, the training process of the Gaussian mixture model is as follows:

[0011] Step 1: Preprocessing and feature extraction: Obtain the relative speed, relative acceleration, and jerk information of the target lane following vehicle with N lane-changing behaviors from the dataset within T time before the ego vehicle changes lanes, and calculate the average relative speed of the target lane following vehicle in the i-th training data. Average relative acceleration and average jerk Recorded as All N training data that meet the conditions are expressed as: P = [P1, P2, ..., P n ];

[0012] Step 2: Parameter training of Gaussian mixture model: All N qualified training data are passed into the Gaussian mixture model. The number of sub-Gaussian models K in the Gaussian mixture model is set to 3. The maximum expectation algorithm is used to solve the parameters of the Gaussian mixture model. It is mainly divided into two steps:

[0013] The first step, E-step, calculates the Q function according to the following formula, which represents the probability that the i-th data belongs to the k-th sub-Gaussian distribution:

[0014]

[0015] Among them, φ k represents the probability of the kth sub-Gaussian distribution, x (i) represents the i-th data, μ k represents the mean of the kth sub-Gaussian distribution, σ k Represents the standard deviation of the k-th sub-Gaussian distribution, and φ is randomly given in the first round of iteration k 、μk , σ k The value of

[0016] The second step, M-step, updates the model parameters according to the following formula:

[0017]

[0018] Repeat the E-step and M-step iterations several times until the model converges.

[0019] After the training is completed, the Gaussian mixture model evaluates the probability x that the following vehicle's driving style is cautious by inputting the following vehicle's average relative speed, average relative acceleration, and average jerk information within the time T before the vehicle changes lanes. d1 , the general probability x d2 , the probability of radical type x d3 , and output the above probability as the output of the driving style evaluation submodule.

[0020] Furthermore, the lane change decision module works as follows: kinematic information of the vehicle behind the target lane, kinematic information of the vehicle itself, kinematic information of surrounding vehicles, and driving style quantification data X of the driving style quantification module are combined. d =[x d1 ,x d2 ,x d3 ] is passed to the lane-changing benefit function submodule. The lane-changing benefit function submodule is based on the established benefit function of the ego vehicle SV and the benefit function of the vehicle behind the target lane PV. First, the driving style and dynamic weight storage submodule is queried to see if there is a dynamic weight record of the same driving style quantification data. If there is a record, the dynamic weight record is called as the weights ω1 and ω2 of the PV benefit function. If there is no record, the dynamic weight record is used as the weights ω1 and ω2 of the PV benefit function. d =[x d1 ,x d2 ,x d3 ]Preliminary adjustment of the weights of the PV return function: Then, the SV benefit function result is calculated based on the established SV benefit function, and the PV benefit function result is calculated based on the established PV benefit function of the target lane. The calculated SV benefit function result, PV benefit function result and each weight of the PV benefit function are passed to the decision judgment submodule, and the decision judgment submodule first determines the lane change benefit R of the SV at this time. lane Is it greater than the lane keeping benefit R of staying in the original lane? stay , if the lane-changing benefit of SV is R lane No greater than the lane keeping benefit R of staying in the original lane stay , then the decision of the vehicle SV to keep the lane is given; if the lane change benefit Rlane Greater than the lane keeping benefit R of staying in the original lane stay , the decision judgment submodule is based on the game idea that PV expects to obtain the maximum benefit. By default, PV will take the action of avoiding and slowing down or continuing to drive without avoiding according to the principle of maximizing its own benefits. Based on this, the decision of whether the self-vehicle SV should change lanes or keep lanes is given. The decision is then passed to the decision evaluation submodule of the lane change performance baseline evaluation module, and the weights of the PV benefit function are passed to the weight self-learning submodule. The decision evaluation submodule determines whether the weights of the current PV benefit function have optimization space and outputs the evaluation results to the weight self-learning submodule. The weight self-learning submodule receives the PV benefits from the decision judgment submodule. The weights of the PV benefit function, the evaluation results from the decision evaluation sub-module, and the driving style quantitative data from the driving style quantitative data temporary storage sub-module. If the evaluation results from the decision evaluation sub-module show that there is room for optimization in the weights of the PV benefit function, the weight self-learning sub-module then guides the dynamic adjustment of the weights of the PV benefit function, and stores the new weights and the corresponding driving style quantitative data in the driving style and dynamic weight storage sub-module. Otherwise, the weight self-learning sub-module does not adjust the weights of the PV benefit function, and directly transmits the corresponding weights and the corresponding driving style quantitative data to the driving style and dynamic weight storage sub-module.

[0021] Specifically, at the current time step t, the lane change benefit function submodule calculates the SV benefit function and PV benefit function results at the current time step t based on the kinematic information of the target lane following vehicle, the ego vehicle kinematic information, the kinematic information of the surrounding vehicles at the current time step t obtained from the data set and the current driving style quantification data obtained from the driving style quantification module. The established ego vehicle SV benefit function and PV benefit function are as follows:

[0022] SV return function S priceSV for:

[0023]

[0024] In the above formula, K SV1 , K SV2 , K SV3 、 Both are proportional coefficients, R lane is the lane-changing benefit of the SV when making a lane-changing decision;

[0025] V S (v SV ,v FV1 ) function represents the desired benefit of the ego vehicle SV to ensure its own driving speed by changing lanes, which is expressed as: Among them, v SVis the velocity of the ego vehicle SV at the current time step t, v FV1 is the velocity of FV1 at the current time step t;

[0026] V FV (v FV1 ,v FV2 ) function represents the benefit of the vehicle’s lane-changing action in improving its subsequent following speed, which is expressed as: Among them, v FV1 is the velocity of FV1 at the current time step t, v FV2 is the velocity of FV2 at the current time step t;

[0027] S PV (S SV ,S PV ) function represents the safety benefit based on the longitudinal distance between the ego vehicle SV and the following vehicle PV, which is expressed as:

[0028]

[0029] Among them, S SV is the longitudinal position of SV at the current time step t (with the vehicle's travel direction as the positive direction of the coordinate system), S PV is the longitudinal position of PV at the current time step t; S PV (S SV ,S PV ) is a piecewise function. When the longitudinal distance S between SV and PV is SV -S PV Exceeding the safety threshold S s Provides a constant positive return R sp ; When the longitudinal distance S between SV and PV SV -S PV At the safety threshold S s With the danger threshold S d When the spacing is between, it provides a positive benefit that gradually decreases as the spacing decreases. Until the risk threshold is reached, the profit drops to 0; when the vertical distance S between SV and PV SV -S PV Below the danger threshold S d When a large negative constant return R is provided spd ;

[0030] Function representation The SV behavior decision constraint function based on the longitudinal distance and longitudinal speed between SV and FV1 is expressed as: Among them, S FV1 is the longitudinal position of FV1 at the current time step t. When the longitudinal arrival time of SV and FV1 is When it is less than 3.5 seconds,

[0031] Provides a large negative return R spn1 ; When the longitudinal arrival time of SV and FV1 When the time is greater than or equal to 3.5 seconds, Provides a small positive return R sp1 ;

[0032] Function representation The SV behavior decision constraint function based on the longitudinal distance and longitudinal speed between SV and FV2 is expressed as: Among them, S FV2 is the longitudinal position of FV2 at the current time step t. When the longitudinal arrival time of SV and FV2 is Less than the minimum lane-changing safety following time difference t sp2 =3.5+t lanec hour, Provides a large negative return R spn2 ; When the longitudinal arrival time of SV and FV2 Greater than or equal to the minimum lane-changing safety following time difference t sp2 hour,

[0033] Provides a small positive return R sp2 , where t lanec is the average lane changing time of the driver, t lanec The average lane changing time of drivers is obtained by counting the public data set;

[0034] When the SV does not change lanes and stays in the original lane, it obtains a constant small negative benefit R stay .

[0035] PV return function S pricePV for:

[0036] S pricePV =ω1·V PV (a PV ,v PV )+ω2·S Space (S PV ,S SV )

[0037] In the above formula, ω1 and ω2 represent the weights of the PV profit function and are dynamically variable positive weight coefficients. PV (a PV ,v PV ) function characterizes the required return on PV speed, which is expressed as: Among them, v PV is the velocity of PV at the current time step t, a PVis the acceleration value of PV at the current time step t. When the action of PV is to avoid deceleration, a PV is a negative constant. When the PV action is not to avoid, a PV =0, τ0 is a positive constant; S Space (S PV ,S SV ) function represents the required benefit of PV, which is expressed as: Among them, h0 is a positive constant, S SV is the longitudinal position of SV at the current time step t, S PV is the longitudinal position of PV at the current time step t.

[0038] Specifically, the decision-making process of the decision judgment submodule is as follows: the decision judgment submodule obtains the SV benefit function result and PV benefit function result of the current time step t from the lane change benefit function submodule, and first determines the lane change benefit R of the SV at this time. lane Is it greater than the lane keeping benefit R of staying in the original lane? stay , if the lane-changing benefit of SV is R lane No greater than the lane keeping benefit R of staying in the original lane stay , the decision-making submodule determines that the lane-changing prerequisite is not met at this time, and outputs the decision that the SV should keep the lane; if the lane-changing benefit R lane Greater than the lane keeping benefit R of staying in the original lane stay , the decision judgment submodule determines that the prerequisite for lane changing is met at this time; then the PV avoidance deceleration benefit and the PV non-avoidance benefit are calculated, and the benefit values ​​of the two are compared. It is considered that PV will take the action with the larger benefit value: if the PV avoidance deceleration benefit is greater than or equal to the non-avoidance benefit, then the PV is judged to take the avoidance deceleration action; if the PV avoidance deceleration benefit is less than the non-avoidance benefit, then the PV is judged to take the non-avoidance action. Based on this, a decision is made on whether the ego vehicle SV should change lanes or continue lane keeping at this time: if the PV action is judged to be avoidance deceleration, the output decision of the decision judgment submodule at this time is lane changing; if the PV action is judged to be non-avoidance, then the output decision of the decision judgment submodule at this time is lane keeping. Then the decision judgment submodule transmits the decision to the decision evaluation submodule of the lane changing performance baseline evaluation module, and transmits the weights of the PV benefit function to the weight self-learning submodule.

[0039] Furthermore, the working process of the lane-changing performance baseline evaluation module is as follows: first, the kinematic information of the target lane vehicle, the kinematic information of the vehicle itself, the kinematic information of the surrounding vehicles, and the driving style quantification data of the driving style quantification module extracted from the data set are used as the current state, and after data preprocessing, are transmitted to the trained data-driven lane-changing decision submodule. The trained data-driven lane-changing decision submodule outputs the action of the current SV, i.e., lane changing or lane keeping, based on the current state input, and then transmits the action to the decision evaluation submodule. The decision evaluation submodule evaluates the lane-changing decision submodule based on the decision judgment submodule of the lane-changing decision module. The module inputs the SV's current decision on whether to change lanes or keep lane, and compares it with the output of the data-driven lane-changing decision submodule to evaluate which of the output decision results of the current data-driven lane-changing decision submodule and the output decision results of the decision judgment submodule of the lane-changing decision module is better. The evaluation results are then output to the baseline self-evaluator and the weighted self-learning submodule of the lane-changing decision module. The weighted self-learning submodule dynamically adjusts the weights of the PV benefit function based on the evaluation results of the decision evaluation submodule. The baseline self-evaluator decides whether the current weighted self-evolution process can be terminated based on the evaluation results of the decision evaluation submodule.

[0040] Specifically, the data preprocessing process before being passed to the trained data-driven lane change decision submodule is as follows:

[0041] The target lane rear vehicle kinematic information, the vehicle kinematic information, the surrounding vehicle kinematic information, and the driving style quantification data X obtained from the driving style quantification module are combined into a data set. d Composition state data set S D :

[0042]

[0043] Among them, i represents that the state is the i-th time series sequence extracted from the data set, 0≤i≤N E ; For the i-th time sequence state sequence S D(i) , expressed as:

[0044]

[0045] Where t represents the current state data, which is the state data of the t-th sampling point, 0≤t≤T E , the i-th time series S D(i,t) The state data at the tth sampling point is expressed as:

[0046] S D(i,t) =[E PV(i,t) ,E SV(i,t) ,E FV1(i,t) ,E FV2(i,t) ,X d(i) ]

[0047] wherein E PV(i,t) represents the kinematic information of the target lane rear vehicle PV at the t-th sampling point, E SV(i,t) represents the kinematic information of the ego vehicle SV at the t-th sampling point, E FV1(i,t) represents the kinematic information of the ego vehicle front vehicle FV1 in the lane before lane change at the t-th sampling point, E FV2(i,t) represents the kinematic information of the target lane front vehicle FV2 at the t-th sampling point, X d(i) represents the driving style quantitative data of the PV in the i-th time sequence.

[0048] Specifically, the training process of the data-driven lane change decision submodule is as follows:

[0049] First, a time sequence state sequence S D is randomly taken from the state data set S D(i) , and then the state data of each sampling point of the time sequence state sequence is sequentially transmitted into the data-driven lane change decision submodule in time sequence order; at the t-th time step, S D(i,t) is transmitted, the deep Q network in the data-driven lane change decision submodule calculates the current action state value Q and makes an action a t , wherein the action a t is one of the two discrete actions of lane keeping or lane changing, is the model parameter of the deep Q network, and then at the t+1-th time step, S D(i,t+1) is transmitted, the deep Q network in the data-driven lane change decision submodule obtains the real reward r S defined by the self-return function R t (t) of the data-driven lane change decision submodule, based on the principle of maximizing the action state value Q at this time, the action a t+1 = argmax Q(s t+1 , a, ω) is calculated, and the time sequence target is obtained. wherein 0 < γ < 1, and the loss Loss is minimized: The model parameter is updated by gradient descent: wherein η is a positive constant;

[0050] The above steps are repeated until the model parameter converges, and the data-driven lane change decision submodule is trained. After the training is completed, S D(i,t) is transmitted into the data-driven lane change decision submodule at the time step t, and the data-driven lane change decision submodule outputs the decision at the time step t.

[0051] Further, the return function R S(t) is defined as follows:

[0052]

[0053] Among them, μ1, μ2, μ3 are constant coefficients, v SV (t) indicates the time from E SV(i,t) The obtained SV longitudinal velocity information, v FV1 (t) indicates the time from E FV1(i,t) The longitudinal velocity information of FV1 is obtained, v FV2 (t) indicates the time from E FV2(i,t) The longitudinal velocity information of FV2 is obtained, S SV (t) indicates the time from E SV(i,t) The longitudinal position information of the SV obtained, S PV (t) indicates the time from E PV(i,t) The longitudinal position information of PV obtained, S d is the danger threshold.

[0054] Specifically, the evaluation process of the decision evaluation submodule is as follows:

[0055] The decision evaluation submodule receives the decision action A1 from the decision judgment submodule of the lane change decision module and the decision action A2 from the data-driven lane change decision submodule. Based on the output values ​​of these two actions, it determines the quality of the action according to the following judgment rules and takes action:

[0056] (1) If A1 = A2 = lane keeping, then at this moment, since lane changing behavior has not occurred, the decision evaluation submodule only outputs decision-consistent information to the baseline self-evaluator.

[0057] (2) If A1 = lane keeping and A2 = lane changing, then the decision evaluation submodule uses the cubic polynomial method to keep the speed at the start and completion of the lane change consistent, and the lane change time is t lanec Plan a lane change route and attempt a lane change; during the lane change attempt, the trajectories of surrounding vehicles and PV are updated based on the kinematic information obtained from the dataset, and the SV's trajectory is updated based on the planned lane change route: If the longitudinal distance between SV and PV falls below the safety threshold S during the lane change attempt, d , it is considered that lane changing at this moment is inferior to lane keeping, so the decision evaluation submodule only outputs the information that is better than the lane changing decision module to the baseline self-evaluator; if the longitudinal distance between SV and PV is always higher than the safety threshold S during the lane changing attempt, d , it is considered that lane changing is better than lane keeping at this moment. Therefore, the decision evaluation submodule outputs the information that lane changing is currently possible to the weight self-learning submodule of the lane changing decision module, and outputs the information of decision inconsistency to the baseline self-evaluator.

[0058] (3) If A1 = lane change and A2 = lane change, then at this moment, since the actions are consistent, the decision evaluation submodule only outputs the decision consistency information to the baseline self-evaluator.

[0059] (4) If A1 = lane change and A2 = lane keeping, then the decision evaluation submodule uses a cubic polynomial to plan the lane change route and attempts a lane change. During the lane change attempt, the trajectories of the surrounding vehicles and the PV are updated based on the kinematic information obtained from the dataset, and the SV's trajectory is updated based on the planned lane change route. If the longitudinal distance between the SV and the PV falls below the safety threshold S during the lane change attempt, d , it is considered that lane changing at this moment is inferior to lane keeping. Therefore, the decision evaluation submodule outputs information about the risk of the current lane change to the weight self-learning submodule of the lane change decision module, and outputs information about the inconsistent decision to the baseline self-evaluator. If the longitudinal distance between SV and PV is always higher than the safety threshold S during the lane change attempt, d , it is considered that lane changing is better than lane keeping at this moment, so the decision evaluation submodule only outputs the information that the lane changing decision module is better to the baseline self-evaluator.

[0060] Specifically, the operation process of the baseline self-assessor is as follows:

[0061] The baseline self-evaluator receives the information output by the decision evaluation submodule and counts the number of information h1 that is better than the lane change decision module, the number of information h2 that is inconsistent with the decision, and the number of information h3 that is consistent with the decision every M (M is a positive constant) data. If and It is considered that the decision of the decision judgment submodule is significantly better than the decision of the data-driven lane change decision submodule at this time, and the weight self-evolution process is terminated; otherwise, the weight self-evolution process continues; where β0 and β1 are positive constants greater than zero.

[0062] Specifically, the operation process of the weight self-learning submodule is as follows:

[0063] The weight self-learning submodule receives the weights of the PV benefit function of the decision judgment submodule and the information output by the decision evaluation submodule:

[0064] (1) When receiving the information from the decision evaluation submodule that the current lane change is risky, it is considered that the weight of the speed benefit term in the PV benefit function is underestimated. Therefore, the weight self-learning submodule adjusts the PV benefit function S according to the following formula: pricePV The weights ω1 and ω2 are the new weights ω1' and ω'2:

[0065] 1-ω1'→ω'2, where α is a positive constant coefficient.

[0066] (2) When receiving the information that the lane can be changed from the decision evaluation submodule, it is considered that the weight of the distance benefit term in the PV benefit function is underestimated. Therefore, the weight self-learning submodule adjusts the PV benefit function S according to the following formula pricePV The weights ω1 and ω2 are the new weights ω1' and ω'2:

[0067] 1-ω'2→ω1', where α is a positive constant coefficient.

[0068] (3) When no information is received from the decision evaluation submodule, the weight self-learning submodule does not adjust the weights of the PV profit function.

[0069] Subsequently, the weight self-learning submodule transfers the latest PV benefit function weights and their corresponding driving style quantitative data to the driving style and dynamic weight storage submodule for storage.

[0070] Specifically, the operation process of the driving style and dynamic weight storage submodule is as follows:

[0071] First, the driving style and dynamic weight storage submodule receives the newly passed PV reward function weights and their corresponding driving style quantitative data from the weight self-learning submodule. Then, the driving style and dynamic weight storage submodule queries the current storage for driving style quantitative data records that match the newly passed driving style quantitative data and performs the following operations:

[0072] (1) If there is a driving style quantification data record that is consistent with the newly input driving style quantification data, the driving style and dynamic weight storage submodule updates the original record and uses the newly input PV benefit function weights to overwrite the original PV benefit function weight records.

[0073] (2) If there is no driving style quantitative data record that is consistent with the newly-input driving style quantitative data, the driving style and dynamic weight storage submodule stores the weights of the newly-input PV benefit function and its corresponding driving style quantitative data as a record in the storage.

[0074] The overall process of weight self-evolution learning of a weight self-evolution intelligent vehicle personalized lane change decision system provided by the present invention is as follows:

[0075] Step one, extract the target lane rear vehicle kinematics information, ego vehicle kinematics information and surrounding vehicle kinematics information in the data set, and then input the above information into the driving style quantification module, lane changing decision module and lane changing performance baseline evaluation module; first, a driving information is randomly extracted from the data set, and the driving information includes the target lane rear vehicle kinematics information, ego vehicle kinematics information and surrounding vehicle kinematics information of each time step in a lane changing process; at time step t, the information input into the driving style quantification module is the kinematics information of the target lane rear vehicle in the data set during T time steps from time step t-T+1 to time step t; at time step t, the information input into the lane changing decision module and the lane changing performance baseline evaluation module is the target lane rear vehicle kinematics information, ego vehicle kinematics information and surrounding vehicle kinematics information of the data set at time step t.

[0076] Step two, the driving style quantification module first inputs the target lane rear vehicle kinematics information into the kinematics feature extraction submodule, the kinematics feature extraction submodule performs data preprocessing and data feature extraction on the input kinematics information, and then inputs the processed information into the driving style evaluation submodule, the trained Gaussian mixture model in the driving style evaluation submodule receives the above information, and gives the driving style quantification probability of the target lane rear vehicle according to the obtained information, including: the probability of cautious type x d1 , the probability of general type x d2 , and the probability of aggressive type x d3 , and the probabilities x d1 , x d2 , x d3 are output to the driving style quantification data temporary storage submodule, the driving style quantification data temporary storage submodule arranges the data into a one-row three-column vector X d =[x d1 ,x d2 ,x d3 ], that is, driving style quantification data, and stores it in the data temporary storage; when new driving style quantification data is transmitted to the driving style quantification data temporary storage submodule, the previous data stored in the data temporary storage will be overwritten by the new quantification data.

[0077] Step three, the lane changing decision module calculates the benefit functions of SV and PV through the lane changing benefit function submodule according to the target lane rear vehicle kinematics information, ego vehicle kinematics information and surrounding vehicle kinematics information, and the driving style quantification data of the driving style quantification module, and outputs the calculation results and the weights of the PV benefit function to the decision judgment submodule; the decision judgment submodule outputs the decision judgment result to the decision evaluation submodule of the lane changing performance baseline evaluation module, and outputs the weights of the PV benefit function to the weight self-learning submodule.

[0078] In step 4, the data-driven lane change decision submodule of the lane change performance baseline assessment module outputs the decision result to the decision assessment submodule based on the kinematic information of the vehicle behind the target lane, the kinematic information of the vehicle itself, the kinematic information of the surrounding vehicles, and the driving style quantification data from the driving style quantification module.

[0079] In step 5, the decision evaluation submodule evaluates whether the output decision result of the current data-driven lane change decision submodule or the output decision result of the decision judgment submodule of the lane change decision module is better, and then outputs the evaluation result to the weight self-learning submodule and baseline self-evaluator of the lane change decision module.

[0080] In step 6, the weight self-learning submodule dynamically adjusts the weights of the PV benefit function based on the evaluation results of the decision evaluation submodule, and transmits the latest weights of the PV benefit function and the corresponding driving style quantification data obtained from the driving style quantification data temporary storage submodule to the driving style and dynamic weight storage submodule of the lane change decision module. The driving style and dynamic weight storage submodule stores the transmitted information.

[0081] In step 7, the baseline self-evaluator decides whether the current weight self-evolution process can be terminated based on the evaluation results of the decision evaluation submodule. If the weight has not yet been adjusted to the optimal state, steps 1 to 7 are repeated; if the weight has been adjusted to the optimal state, the weight self-evolution process is terminated.

[0082] Beneficial effects of the present invention:

[0083] 1. The weighted, self-evolving, personalized lane-changing decision system for intelligent vehicles described in this invention analyzes the explicit kinematic characteristics of a following vehicle in an adjacent lane when the ego vehicle initiates a lane-changing attempt. This system abstractly characterizes the relationship between the kinematic parameters of the following vehicle and the driving style of the following vehicle's driver. Furthermore, it establishes a driving style quantification index, using a probabilistic model to quantitatively describe the driving style of the following vehicle driver and the differences between different driving styles.

[0084] 2. The weighted, self-evolving, personalized lane-changing decision system for intelligent vehicles described in this invention deeply considers the behavioral interactions between vehicles during actual lane changes and establishes a lane-changing decision module based on game theory. This allows the system to explicitly represent the dynamic, time-varying interactions between vehicles during lane changes, thereby making the lane-changing decision logic clear and transparent, and enhancing the interpretability of the algorithm. By comprehensively considering multiple driving benefits, a practical, multi-weighted game payoff function is established. Lane-changing decisions are made based on this payoff function, improving the stability of the system's decision output.

[0085] 3. The weighted self-evolving personalized lane-changing decision system for intelligent vehicles described in the present invention significantly improves the adaptability and flexibility of the lane-changing decision system by dynamically adjusting the weights of the game payoff function. It introduces a lane-changing performance baseline evaluation module based on deep reinforcement learning and couples it with a driving style quantification module, thereby establishing a quantitative mapping relationship between driving style and lane-changing decision results. The established lane-changing decision system is then equipped with autonomous learning capabilities, enabling it to achieve good performance in various complex and changeable driving conditions, making it possible to guide the stable improvement of the lane-changing decision system through a data-driven approach. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 Schematic diagram of the overall framework of the lane change decision system of the present invention.

[0087] Figure 2 Schematic diagram of the positional relationship between the ego vehicle SV and surrounding traffic vehicles in a typical lane-changing scenario.

[0088] Figure 3 This is a schematic diagram of the driving style quantification module principle of the present invention.

[0089] Figure 4 This is a flow chart of the decision-making and judgment submodule of the present invention.

[0090] Figure 5 Schematic diagram of the framework of the data-driven lane change decision submodule of the present invention.

[0091] Figure 6 Schematic diagram of the actual application flow of the decision-making system of the present invention. DETAILED DESCRIPTION

[0092] See Figure 1 As shown, the present invention provides a weighted self-evolving intelligent vehicle personalized lane-changing decision system, which includes three main modules: a driving style quantification module, a lane-changing decision module, and a lane-changing performance baseline assessment module. The driving style quantification module includes a kinematic feature extraction submodule, a driving style assessment submodule, and a driving style quantification data temporary storage submodule. The lane-changing decision module includes a lane-changing benefit function submodule, a decision judgment submodule, a weighted self-learning submodule, and a driving style and dynamic weight storage submodule. The lane-changing performance baseline assessment module includes a data-driven lane-changing decision submodule, a decision assessment submodule, and a baseline self-assessor.

[0093] First, a driving record is randomly extracted from the dataset. A driving record includes the kinematic information of the target lane's following vehicle, the ego vehicle's kinematic information, and the kinematic information of surrounding vehicles at each time step of a lane change process. At time step t, the information input to the driving style quantification module is the kinematic information of the target lane's following vehicle during the time T from time step t-T+1 to time step t. At time step t, the information input to the lane change decision module and the lane change performance baseline assessment module is the kinematic information of the target lane's following vehicle, the ego vehicle's kinematic information, and the kinematic information of surrounding vehicles at time step t.

[0094] The target lane's following vehicle kinematic information includes the target lane's following vehicle (PV)'s longitudinal and lateral position, longitudinal and lateral velocity, longitudinal and lateral acceleration, and jerk, as well as its relative velocity and acceleration to the ego vehicle (SV). The ego vehicle's kinematic information includes the ego vehicle's (SV) longitudinal and lateral position, longitudinal and lateral velocity, and longitudinal and lateral acceleration. The surrounding vehicle kinematic information includes the longitudinal and lateral position and longitudinal and lateral velocity of the preceding vehicle (FV1) in the lane the ego vehicle was in before the lane change and the preceding vehicle (FV2) in the target lane. All of this information is obtained from public datasets.

[0095] Typical lane-changing scenarios include Figure 2 As shown in the figure, SV is the lane-changing ego vehicle, PV is the following vehicle in the target lane, and SV intends to change lanes from its current lane to PV's lane. The preceding vehicle FV1 in the lane SV is in before changing lanes and the preceding vehicle FV2 in the target lane provide kinematic constraints for SV's lane-changing behavior. The interactive confrontation between SV and PV determines SV's lane-changing decision.

[0096] like Figure 1 、 3 As shown, the working process of the driving style quantification module is as follows: first, the kinematic information of the vehicle behind the target lane is transmitted to the kinematic feature extraction submodule, which performs data preprocessing and data feature extraction on the transmitted kinematic information. The processed information is then transmitted to the driving style evaluation submodule. The trained Gaussian mixture model (GMM) in the driving style evaluation submodule receives the above information and, based on the information obtained, gives the quantified probability of each driving style of the vehicle behind the target lane, including: the probability of cautious type x d1 , the general probability x d2 , the probability of radical type x d3 , and the probability x d1 、x d2 、x d3 Output to the driving style quantization data temporary storage submodule, which organizes the data into a vector X with one row and three columns d =[x d1 ,x d2x d3 ], namely driving style quantification data, and is stored in a data buffer; when new driving style quantification data is transmitted to the driving style quantification data buffer sub-module, the previous data stored in the data buffer will be overwritten by the new quantification data.

[0097] Specifically, the data feature extraction function of the kinematic feature extraction sub-module is to obtain the relative speed, relative acceleration and jerk information of the target lane rear vehicle relative to the ego vehicle within T time, calculate the average relative speed, average relative acceleration and average jerk, and transmit the calculated average relative speed, average relative acceleration and average jerk information to the driving style evaluation sub-module.

[0098] Specifically, the training process of the Gaussian mixture model is as follows:

[0099] Step 1, preprocessing and feature extraction: obtain the relative speed, relative acceleration and jerk information of the target lane rear vehicle relative to the ego vehicle within T time before the lane change of the target lane rear vehicle in the data set, calculate the average relative speed average relative acceleration and average jerk of the target lane rear vehicle of the i-th training data, and record them as All N training data meeting the conditions are expressed as: P = [P1, P2,..., P n ];

[0100] Step 2, parameter training of Gaussian mixture model: transmit all N training data meeting the conditions to the Gaussian mixture model, set the number of sub-Gaussian models K in the Gaussian mixture model to 3, and solve the parameters of the Gaussian mixture model using the expectation maximization algorithm, which mainly includes two steps:

[0101] First step, E-step, calculate the Q function according to the following formula, which represents the probability that the i-th data belongs to the k-th sub-Gaussian distribution:

[0102]

[0103] where φ k represents the probability of the k-th sub-Gaussian distribution, x (i) represents the i-th data, μ k represents the mean of the k-th sub-Gaussian distribution, σ k represents the standard deviation of the k-th sub-Gaussian distribution, and the values of φ k , μ k , σ k are randomly given in the first round of iteration;

[0104] Second step, M-step, update the parameters of the model according to the following formula:

[0105]

[0106] Repeat the E-step and M-step iterations several times until the model converges.

[0107] After the training is completed, the Gaussian mixture model evaluates the probability x that the following vehicle's driving style is cautious by inputting the following vehicle's average relative speed, average relative acceleration, and average jerk information within the time T before the vehicle changes lanes. d1 , the general probability x d2 , the probability of radical type x d3 , and output the above probability as the output of the driving style evaluation submodule.

[0108] like Figure 1 、 4 As shown, the working process of the lane change decision module is as follows: the kinematic information of the vehicle behind the target lane, the kinematic information of the vehicle itself, the kinematic information of the surrounding vehicles, and the driving style quantification data X of the driving style quantification module are combined. d =[x d1 ,x d2 ,x d3 ] is passed to the lane-changing benefit function submodule. The lane-changing benefit function submodule is based on the established benefit function of the ego vehicle SV and the benefit function of the vehicle behind the target lane PV. First, the driving style and dynamic weight storage submodule is queried to see if there is a dynamic weight record of the same driving style quantification data. If there is a record, the dynamic weight record is called as the weights ω1 and ω2 of the PV benefit function. If there is no record, the dynamic weight record is used as the weights ω1 and ω2 of the PV benefit function. d =[x d1 ,x d2 ,x d3 ]Preliminary adjustment of the weights of the PV return function: Then, the SV benefit function result is calculated based on the established SV benefit function, and the PV benefit function result is calculated based on the established PV benefit function of the target lane. The calculated SV benefit function result, PV benefit function result and each weight of the PV benefit function are passed to the decision judgment submodule, and the decision judgment submodule first determines the lane change benefit R of the SV at this time. lane Is it greater than the lane keeping benefit R of staying in the original lane? stay , if the lane-changing benefit of SV is R lane No greater than the lane keeping benefit R of staying in the original lane stay , then the decision of the vehicle SV to keep the lane is given; if the lane change benefit R lane Greater than the lane keeping benefit R of staying in the original lane stay, the decision judgment submodule is based on the game idea that PV expects to obtain the maximum benefit. By default, PV will take the action of avoiding and slowing down or continuing to drive without avoiding according to the principle of maximizing its own benefits. Based on this, the decision of whether the self-vehicle SV should change lanes or keep lanes is given. The decision is then passed to the decision evaluation submodule of the lane change performance baseline evaluation module, and the weights of the PV benefit function are passed to the weight self-learning submodule. The decision evaluation submodule determines whether the weights of the current PV benefit function have optimization space and outputs the evaluation results to the weight self-learning submodule. The weight self-learning submodule receives the PV benefits from the decision judgment submodule. The weights of the PV benefit function, the evaluation results from the decision evaluation sub-module, and the driving style quantitative data from the driving style quantitative data temporary storage sub-module. If the evaluation results from the decision evaluation sub-module show that there is room for optimization in the weights of the PV benefit function, the weight self-learning sub-module then guides the dynamic adjustment of the weights of the PV benefit function, and stores the new weights and the corresponding driving style quantitative data in the driving style and dynamic weight storage sub-module. Otherwise, the weight self-learning sub-module does not adjust the weights of the PV benefit function, and directly transmits the corresponding weights and the corresponding driving style quantitative data to the driving style and dynamic weight storage sub-module.

[0109] Specifically, at the current time step t, the lane change benefit function submodule calculates the SV benefit function and PV benefit function results at the current time step t based on the kinematic information of the target lane following vehicle, the ego vehicle kinematic information, the kinematic information of the surrounding vehicles at the current time step t obtained from the data set and the current driving style quantification data obtained from the driving style quantification module. The established ego vehicle SV benefit function and PV benefit function are as follows:

[0110] SV return function S priceSV for:

[0111]

[0112] In the above formula, K SV1 , K SV2 , K SV3 、 Both are proportional coefficients, R lane is the lane-changing benefit of the SV when making a lane-changing decision;

[0113] V S (v SV ,v FV1 ) function represents the desired benefit of the ego vehicle SV to ensure its own driving speed by changing lanes, which is expressed as: Among them, v SV is the velocity of the ego vehicle SV at the current time step t, v FV1 is the velocity of FV1 at the current time step t;

[0114] V FV (v FV1 ,v FV2 ) function represents the benefit of the vehicle’s lane-changing action in improving its subsequent following speed, which is expressed as: Among them, v FV1 is the velocity of FV1 at the current time step t, v FV2 is the velocity of FV2 at the current time step t;

[0115] S PV (S SV ,S PV ) function represents the safety benefit based on the longitudinal distance between the ego vehicle SV and the following vehicle PV, which is expressed as:

[0116]

[0117] Among them, S SV is the longitudinal position of SV at the current time step t (with the vehicle's travel direction as the positive direction of the coordinate system), S PV is the longitudinal position of PV at the current time step t; S PV (S SV ,S PV ) is a piecewise function. When the longitudinal distance S between SV and PV is SV -S PV Exceeding the safety threshold S s Provides a constant positive return R sp ; When the longitudinal distance S between SV and PV SV -S PV At the safety threshold S s With the danger threshold S d When the spacing is between, it provides a positive benefit that gradually decreases as the spacing decreases. Until the risk threshold is reached, the profit drops to 0; when the vertical distance S between SV and PV SV -S PV Below the danger threshold S d When a large negative constant return R is provided spd ;

[0118] Function representation The SV behavior decision constraint function based on the longitudinal distance and longitudinal speed between SV and FV1 is expressed as: Among them, S FV1 is the longitudinal position of FV1 at the current time step t. When the longitudinal arrival time of SV and FV1 is When it is less than 3.5 seconds, Provides a large negative return R spn1 ; When the longitudinal arrival time of SV and FV1 when the longitudinal distance between the SV and the FV2 is greater than or equal to 3.5 seconds, a small positive value of R is provided sp1 ;

[0119] The function represents the SV behavior decision constraint function according to the longitudinal distance and the longitudinal velocity of the SV and the FV2, and is expressed as: wherein S FV2 is the longitudinal position of the FV2 at the current time step t, and the longitudinal arrival time of the SV and the FV2 is less than the minimum lane-changing safety following time difference t sp2 = 3.5 + t lanec , a large negative value of R is provided spn2 ; when the longitudinal arrival time of the SV and the FV2 is greater than or equal to the minimum lane-changing safety following time difference t sp2 ,

[0120] a small positive value of R is provided sp2 , wherein t lanec is the average lane-changing time of the driver, and t lanec is obtained by statistically analyzing the average lane-changing time of the driver through a public data set;

[0121] when the SV does not perform lane changing, i.e., remains in the original lane, a constant small negative value of R stay is obtained.

[0122] The PV reward function S pricePV is:

[0123] S pricePV = ω1·V PV (a PV ,v PV ) + ω2·S Space (S PV ,S SV )

[0124] In the above formula, ω1 and ω2 represent the weights of each term of the PV reward function, which are positive weight coefficients that can be dynamically changed, V PV (a PV ,v PV ) represents the speed requirement reward of the PV, and is expressed as: wherein v PV is the speed of the PV at the current time step t, and a PV is the acceleration value of the PV at the current time step t, when the action of the PV is to avoid deceleration, a PV is a negative constant, and when the action of the PV is not to avoid, a PV=0, τ0 is a positive constant; S Space (S PV ,S SV ) function represents the required benefit of PV, which is expressed as: Among them, h0 is a positive constant, S SV is the longitudinal position of SV at the current time step t, S PV is the longitudinal position of PV at the current time step t.

[0125] Specifically, the decision-making process of the decision judgment submodule is as follows: the decision judgment submodule obtains the SV benefit function result and PV benefit function result of the current time step t from the lane change benefit function submodule, and first determines the lane change benefit R of the SV at this time. lane Is it greater than the lane keeping benefit R of staying in the original lane? stay , if the lane-changing benefit of SV is R lane No greater than the lane keeping benefit R of staying in the original lane stay , the decision-making submodule determines that the lane-changing prerequisite is not met at this time, and outputs the decision that the SV should keep the lane; if the lane-changing benefit R lane Greater than the lane keeping benefit R of staying in the original lane stay , the decision judgment submodule determines that the prerequisite for lane changing is met at this time; then the PV avoidance deceleration benefit and the PV non-avoidance benefit are calculated, and the benefit values ​​of the two are compared. It is considered that PV will take the action with the larger benefit value: if the PV avoidance deceleration benefit is greater than or equal to the non-avoidance benefit, then the PV is judged to take the avoidance deceleration action; if the PV avoidance deceleration benefit is less than the non-avoidance benefit, then the PV is judged to take the non-avoidance action. Based on this, a decision is made on whether the ego vehicle SV should change lanes or continue lane keeping at this time: if the PV action is judged to be avoidance deceleration, the output decision of the decision judgment submodule at this time is lane changing; if the PV action is judged to be non-avoidance, then the output decision of the decision judgment submodule at this time is lane keeping. Then the decision judgment submodule transmits the decision to the decision evaluation submodule of the lane changing performance baseline evaluation module, and transmits the weights of the PV benefit function to the weight self-learning submodule.

[0126] like Figure 1 、 25, the working process of the lane changing performance baseline evaluation module is as follows: first, the target lane rear vehicle kinematic information, the vehicle kinematic information, the surrounding vehicle kinematic information, and the driving style quantification data of the driving style quantification module extracted from the data set are used as the current state, and after data preprocessing, they are transmitted to the trained data-driven lane changing decision submodule. The trained data-driven lane changing decision submodule outputs the action of the current SV, i.e., lane changing or lane keeping, based on the current state transmitted, and then transmits the action to the decision evaluation submodule. The decision evaluation submodule judges the lane changing decision submodule based on the decision judgment submodule of the lane changing decision module. The module inputs the SV's current decision on whether to change lanes or keep lane, and compares it with the output of the data-driven lane-changing decision submodule to evaluate which of the output decision results of the current data-driven lane-changing decision submodule and the output decision results of the decision judgment submodule of the lane-changing decision module is better. The evaluation results are then output to the baseline self-evaluator and the weighted self-learning submodule of the lane-changing decision module. The weighted self-learning submodule dynamically adjusts the weights of the PV benefit function based on the evaluation results of the decision evaluation submodule. The baseline self-evaluator decides whether the current weighted self-evolution process can be terminated based on the evaluation results of the decision evaluation submodule.

[0127] Specifically, the data preprocessing process before being passed to the trained data-driven lane change decision submodule is as follows:

[0128] The target lane rear vehicle kinematic information, the vehicle kinematic information, the surrounding vehicle kinematic information, and the driving style quantification data X obtained from the driving style quantification module are combined into a data set. d Composition state data set S D :

[0129]

[0130] Among them, i represents that the state is the i-th time series sequence extracted from the data set, 0≤i≤N E ; For the i-th time sequence state sequence S D(i) , expressed as:

[0131]

[0132] Where t represents the current state data, which is the state data of the t-th sampling point, 0≤t≤T E , the i-th time series S D(i,t) The state data at the tth sampling point is expressed as:

[0133] S D(i,t) =[E PV(i,t) ,E SV(i,t) ,E FV1(i,t) ,E FV2(i,t) ,X d(i) ]

[0134] wherein E PV(i,t) represents the kinematic information of the target lane rear vehicle PV at the t-th sampling point, E SV(i,t) represents the kinematic information of the ego vehicle SV at the t-th sampling point, E FV1(i,t) represents the kinematic information of the ego vehicle front vehicle FV1 in the lane before lane change at the t-th sampling point, E FV2(i,t) represents the kinematic information of the target lane front vehicle FV2 at the t-th sampling point, X d(i) represents the driving style quantitative data of the PV in the i-th time sequence.

[0135] Specifically, the training process of the data-driven lane change decision submodule is as follows:

[0136] First, a time sequence state sequence S D is randomly taken from the state data set S D(i) , and then the state data of each sampling point of the time sequence state sequence is sequentially transmitted into the data-driven lane change decision submodule in time sequence order; at the t-th time step, S D(i,t) is transmitted, the deep Q network in the data-driven lane change decision submodule calculates the current action state value Q and makes an action a t , wherein the action a t is one of the two discrete actions of lane keeping or lane changing, is the model parameter of the deep Q network, and then at the t+1-th time step, S D(i,t+1) is transmitted, the deep Q network in the data-driven lane change decision submodule obtains the real reward r S defined by the self-return function R t (t) of the data-driven lane change decision submodule, based on the principle of maximizing the action state value Q at this time, the time sequence target is calculated a t+1 = argmax Q (s t+1 , a, ω), wherein 0 < γ < 1, and the loss Loss is minimized: the model parameter is updated by gradient descent: wherein η is a positive constant;

[0137] The above steps are repeated until the model parameter converges, and the data-driven lane change decision submodule is trained. After the training is completed, S D(i,t) is transmitted into the data-driven lane change decision submodule at the time step t, and the data-driven lane change decision submodule outputs the decision at the time step t.

[0138] Further, the return function R S(t) is defined as follows:

[0139]

[0140] where μ1, μ2, μ3 are constant coefficients, v SV (t) represents the longitudinal velocity information of the SV acquired from the E SV(i,t) (t) represents the longitudinal velocity information of the FV1 acquired from the E FV1 (t) represents the longitudinal velocity information of the FV2 acquired from the E FV1(i,t) (t) represents the longitudinal position information of the SV acquired from the E FV2 (t) represents the longitudinal position information of the PV acquired from the E FV2(i,t) (t) represents the longitudinal position information of the PV acquired from the E SV (t) represents the longitudinal position information of the SV acquired from the E SV(i,t) (t) represents the longitudinal position information of the SV acquired from the E PV (t) represents the longitudinal position information of the PV acquired from the E PV(i,t) (t) represents the longitudinal position information of the PV acquired from the E d is a safety threshold.

[0141] Specifically, the evaluation process of the decision evaluation submodule is as follows:

[0142] The decision evaluation submodule receives the decision action A1 (i.e., SV lane change or lane keeping) from the decision judgment submodule of the lane change decision module and the decision action A2 (i.e., SV lane change or lane keeping) from the data-driven lane change decision submodule, and determines the pros and cons of the actions according to the following judgment rules and actions based on the output values of the two actions:

[0143] (1) If A1 = A2 = lane keeping, no lane change behavior occurs at this moment, and the decision evaluation submodule only outputs decision consistent information to the baseline self-evaluator.

[0144] (2) If A1 = lane keeping and A2 = lane change, the decision evaluation submodule uses a cubic polynomial method to keep the speed consistent at the lane change start time and the lane change completion time, and the lane change duration is t lanec The planned lane change route is used to attempt lane change, and the cubic polynomial path planning is a common method for path planning, the specific details of which are not described here. In the process of attempting lane change, the driving trajectories of surrounding vehicles and the PV are updated based on the kinematic information obtained from the data set, and the driving trajectory of the SV is updated based on the planned lane change route: if the longitudinal distance between the SV and the PV is lower than the safety threshold S d during the process of attempting lane change, it is considered that lane change is inferior to lane keeping at this moment, and therefore the decision evaluation submodule only outputs the more optimal information of the lane change decision module to the baseline self-evaluator; if the longitudinal distance between the SV and the PV is always higher than the safety threshold S d, it is considered that lane changing is better than lane keeping at this moment. Therefore, the decision evaluation submodule outputs the information that lane changing is currently possible to the weight self-learning submodule of the lane changing decision module, and outputs the information of decision inconsistency to the baseline self-evaluator.

[0145] (3) If A1 = lane change and A2 = lane change, then at this moment, since the actions are consistent, the decision evaluation submodule only outputs the decision consistency information to the baseline self-evaluator.

[0146] (4) If A1 = lane change and A2 = lane keeping, then the decision evaluation submodule uses a cubic polynomial to plan the lane change route and attempts a lane change. During the lane change attempt, the trajectories of the surrounding vehicles and the PV are updated based on the kinematic information obtained from the dataset. The SV's trajectory is no longer updated based on the kinematic information obtained from the dataset, but is updated based on the planned lane change route. If the longitudinal distance between the SV and the PV falls below the safety threshold S during the lane change attempt, the SV's trajectory is updated based on the kinematic information obtained from the dataset. d , it is considered that lane changing at this moment is inferior to lane keeping. Therefore, the decision evaluation submodule outputs information about the risk of the current lane change to the weight self-learning submodule of the lane change decision module, and outputs information about the inconsistent decision to the baseline self-evaluator. If the longitudinal distance between SV and PV is always higher than the safety threshold S during the lane change attempt, d , it is considered that lane changing is better than lane keeping at this moment, so the decision evaluation submodule only outputs the information that the lane changing decision module is better to the baseline self-evaluator.

[0147] Specifically, the operation process of the baseline self-assessor is as follows:

[0148] The baseline self-evaluator receives the information output by the decision evaluation submodule and counts the number of information h1 that is better than the lane change decision module, the number of information h2 that is inconsistent with the decision, and the number of information h3 that is consistent with the decision every M (M is a positive constant) data. If and It is considered that the decision of the decision judgment submodule is significantly better than the decision of the data-driven lane change decision submodule at this time, and the weight self-evolution process is terminated; otherwise, the weight self-evolution process continues; where β0 and β1 are positive constants greater than zero.

[0149] Specifically, the operation process of the weight self-learning submodule is as follows:

[0150] The weight self-learning submodule receives the weights of the PV benefit function of the decision judgment submodule and the information output by the decision evaluation submodule:

[0151] (1) When receiving the information from the decision evaluation submodule that the current lane change is risky, it is considered that the weight of the speed benefit term in the PV benefit function is underestimated. Therefore, the weight self-learning submodule adjusts the PV benefit function S according to the following formula: pricePV The weights ω1 and ω2 are the new weights ω1' and ω'2:

[0152] 1-ω1'→ω'2, where α is a positive constant coefficient.

[0153] (2) When receiving the information that the lane can be changed from the decision evaluation submodule, it is considered that the weight of the distance benefit term in the PV benefit function is underestimated. Therefore, the weight self-learning submodule adjusts the PV benefit function S according to the following formula pricePV The weights ω1 and ω2 are the new weights ω1' and ω'2:

[0154] 1-ω'2→ω1', where α is a positive constant coefficient.

[0155] (3) When no information is received from the decision evaluation submodule, the weight self-learning submodule does not adjust the weights of the PV profit function.

[0156] Subsequently, the weight self-learning submodule transfers the latest PV benefit function weights and their corresponding driving style quantitative data to the driving style and dynamic weight storage submodule for storage.

[0157] Specifically, the operation process of the driving style and dynamic weight storage submodule is as follows:

[0158] First, the driving style and dynamic weight storage submodule receives the newly passed PV reward function weights and their corresponding driving style quantitative data from the weight self-learning submodule. Then, the driving style and dynamic weight storage submodule queries the current storage for driving style quantitative data records that match the newly passed driving style quantitative data and performs the following operations:

[0159] (1) If there is a driving style quantification data record that is consistent with the newly input driving style quantification data, the driving style and dynamic weight storage submodule updates the original record and uses the newly input PV benefit function weights to overwrite the original PV benefit function weight records.

[0160] (2) If there is no driving style quantitative data record that is consistent with the newly-input driving style quantitative data, the driving style and dynamic weight storage submodule stores the weights of the newly-input PV benefit function and its corresponding driving style quantitative data as a record in the storage.

[0161] The overall process of weight self-evolution learning of a weight self-evolution intelligent vehicle personalized lane change decision system provided by the present invention is as follows:

[0162] Step 1: Extract the kinematic information of the target lane's following vehicle, the ego vehicle's kinematic information, and the surrounding vehicle's kinematic information from the dataset. This information is then fed into the driving style quantification module, the lane change decision module, and the lane change performance baseline assessment module. First, a driving record is randomly extracted from the dataset. Each driving record includes the kinematic information of the target lane's following vehicle, the ego vehicle's kinematic information, and the surrounding vehicle's kinematic information at each time step of a lane change process. At time step t, the information fed into the driving style quantification module is the kinematic information of the target lane's following vehicle from time step t-T+1 to time step t, a total of T time steps. At time step t, the information fed into the lane change decision module and the lane change performance baseline assessment module is the kinematic information of the target lane's following vehicle, the ego vehicle's kinematic information, and the surrounding vehicle's kinematic information at time step t.

[0163] In step 2, the driving style quantification module first passes the kinematic information of the vehicle behind the target lane to the kinematic feature extraction submodule. The kinematic feature extraction submodule performs data preprocessing and data feature extraction on the incoming kinematic information, and then passes the processed information to the driving style evaluation submodule. The trained Gaussian mixture model in the driving style evaluation submodule receives the above information and, based on the information obtained, gives the quantified probability of each driving style of the vehicle behind the target lane, including: the probability of cautious type x d1 , the general probability x d2 , the probability of radical type x d3 , and the probability x d1 、x d2 、x d3 Output to the driving style quantization data temporary storage submodule, which organizes the data into a vector X with one row and three columns d =[x d1 ,x d2 ,x d3 ], that is, driving style quantization data, and is stored in the data register; when new driving style quantization data is transmitted to the driving style quantization data temporary storage sub-module, the previous data stored in the data register will be overwritten by the new quantization data.

[0164] In step 3, the lane change decision module calculates the benefit functions of SV and PV through the lane change benefit function submodule based on the kinematic information of the vehicle behind the target lane, the kinematic information of the vehicle itself, the kinematic information of the surrounding vehicles, and the driving style quantification data of the driving style quantification module, and outputs the calculation results and the weights of the PV benefit function to the decision judgment submodule; the decision judgment submodule outputs the decision judgment result to the decision evaluation submodule of the lane change performance baseline evaluation module, and outputs the weights of the PV benefit function to the weight self-learning submodule.

[0165] In step 4, the data-driven lane change decision submodule of the lane change performance baseline assessment module outputs the decision result to the decision assessment submodule based on the kinematic information of the vehicle behind the target lane, the kinematic information of the vehicle itself, the kinematic information of the surrounding vehicles, and the driving style quantification data from the driving style quantification module.

[0166] In step 5, the decision evaluation submodule evaluates whether the output decision result of the current data-driven lane change decision submodule or the output decision result of the decision judgment submodule of the lane change decision module is better, and then outputs the evaluation result to the weight self-learning submodule and baseline self-evaluator of the lane change decision module.

[0167] In step 6, the weight self-learning submodule dynamically adjusts the weights of the PV benefit function based on the evaluation results of the decision evaluation submodule, and transmits the latest weights of the PV benefit function and the corresponding driving style quantification data obtained from the driving style quantification data temporary storage submodule to the driving style and dynamic weight storage submodule of the lane change decision module. The driving style and dynamic weight storage submodule stores the transmitted information.

[0168] In step 7, the baseline self-evaluator decides whether the current weight self-evolution process can be terminated based on the evaluation results of the decision evaluation submodule. If the weight has not yet been adjusted to the optimal state, steps 1 to 7 are repeated; if the weight has been adjusted to the optimal state, the weight self-evolution process is terminated.

[0169] See Figure 2 and Figure 6 When the baseline self-evaluator determines that the weight self-learning process can be terminated, it is considered that the weight has been adjusted to the optimal state. At this time, the lane change decision system is trained and has the ability to make lane change decisions in a real traffic environment. The following is a practical application process:

[0170] In step 1, the ego vehicle (SV) uses sensing devices such as lidar and cameras, as well as ego vehicle kinematic monitoring equipment, to obtain kinematic information about the vehicle behind it in the current target lane, its own vehicle, and surrounding vehicles. This information is then fed into the driving style quantification module, the lane change decision module, and the lane change performance baseline assessment module. At time step t, the information fed into the driving style quantification module is the kinematic information of the vehicle behind it in the target lane from time step t-T+1 to time step t, a total of T. At time step t, the information fed into the lane change decision module and the lane change performance baseline assessment module is the kinematic information of the vehicle behind it in the target lane, its own vehicle, and surrounding vehicles at time step t.

[0171] Step 2: The driving style quantification module gives the driving style quantification result X based on the kinematic information of the vehicle behind the target lane. di , output to the lane-changing decision module.

[0172] Step 3: The driving style and dynamic weight storage submodule of the lane change decision module quantifies the driving style result X di , find the weight stored during evolution and the driving style quantization result X di The driving style quantification result X with the smallest Euclidean distance dj , X dj The corresponding dynamic weights and the kinematic information of the vehicle behind the target lane, the kinematic information of the vehicle itself, and the kinematic information of the surrounding vehicles are output to the lane change benefit function submodule.

[0173] Step 4: The lane-changing profit function submodule calculates the profit function of SV and PV and outputs the calculation results to the decision-making submodule.

[0174] In step 5, the decision-making submodule outputs the final decision action, that is, whether to change lanes or keep lanes. If the final decision action is to change lanes, the decision process ends and the lane change signal is sent to the vehicle controller to execute the lane change action. If the final decision action is to keep lanes, steps 1 to 5 are repeated.

Claims

1. A weighted self-evolving personalized lane-changing decision system for intelligent vehicles, characterized by: Includes driving style quantification module, lane change decision module, and lane change performance baseline assessment module; The driving style quantification module includes a kinematic feature extraction submodule, a driving style evaluation submodule, and a driving style quantification data temporary storage submodule; The lane change decision module includes: a lane change benefit function submodule, a decision judgment submodule, a weight self-learning submodule, and a driving style and dynamic weight storage submodule; the lane change performance baseline assessment module includes a data-driven lane change decision submodule, a decision assessment submodule, and a baseline self-assessor; The working process of the driving style quantification module is as follows: First, the kinematic information of the vehicle behind the target lane is transmitted to the kinematic feature extraction submodule, which performs data preprocessing and data feature extraction on the transmitted kinematic information. The processed information is then transmitted to the driving style evaluation submodule. The trained Gaussian mixture model in the driving style evaluation submodule receives the above information and, based on the information obtained, gives the quantified probability of each driving style of the vehicle behind the target lane, including: the probability of cautious type x d1 , the general probability x d2 , the probability of radical type x d3 , and the probability x d1 、x d2 、x d3 Output to the driving style quantization data temporary storage submodule, which organizes the data into a vector X with one row and three columns d =[x d1 ,x d2 ,x d3 ], i.e., driving style quantization data, and is stored in the data register; when new driving style quantization data is transmitted to the driving style quantization data temporary storage submodule, the previous data stored in the data register will be overwritten by the new quantization data; The working process of the lane change decision module is as follows: the kinematic information of the vehicle behind the target lane, the kinematic information of the vehicle itself, the kinematic information of the surrounding vehicles, and the driving style quantification data X of the driving style quantification module are combined. d =[x d1 ,x d2 ,x d3 ] is passed to the lane-changing benefit function submodule. The lane-changing benefit function submodule is based on the established benefit function of the ego vehicle SV and the benefit function of the vehicle behind the target lane PV. First, the driving style and dynamic weight storage submodule is queried to see if there is a dynamic weight record of the same driving style quantification data. If there is a record, the dynamic weight record is called as the weights ω1 and ω2 of the PV benefit function. If there is no record, the dynamic weight record is used as the weights ω1 and ω2 of the PV benefit function. d =[x d1 ,x d2 ,x d3 ]Preliminary adjustment of the weights of the PV return function: Then, the SV benefit function result is calculated based on the established SV benefit function, and the PV benefit function result is calculated based on the established PV benefit function of the target lane. The calculated SV benefit function result, PV benefit function result and each weight of the PV benefit function are passed to the decision judgment submodule, and the decision judgment submodule first determines the lane change benefit R of the SV at this time. lane Is it greater than the lane keeping benefit R of staying in the original lane? stay , if the lane-changing benefit of SV is R lane No greater than the lane keeping benefit R of staying in the original lane stay , then the decision of the vehicle SV to keep the lane is given; if the lane change benefit R lane Greater than the lane keeping benefit R of staying in the original lane stay , the decision judgment submodule is based on the game idea that PV expects to obtain the maximum benefit. By default, PV will take the action of avoiding and slowing down or continuing to drive without avoiding according to the principle of maximizing its own benefits. Based on this, the decision of whether the self-vehicle SV should change lanes or keep lanes is given. The decision is then passed to the decision evaluation submodule of the lane change performance baseline evaluation module, and the weights of the PV benefit function are passed to the weight self-learning submodule. The decision evaluation submodule determines whether the weights of the current PV benefit function have optimization space and outputs the evaluation results to the weight self-learning submodule. The weight self-learning submodule receives the PV benefits from the decision judgment submodule. The weights of the PV-reward function, the evaluation results from the decision-making evaluation submodule, and the driving style quantitative data from the driving style quantitative data temporary storage submodule are collected. If the evaluation results from the decision-making evaluation submodule indicate that there is room for optimization in the weights of the PV-reward function, the weight self-learning submodule then guides the dynamic adjustment of the weights of the PV-reward function and stores the new weights and the corresponding driving style quantitative data in the driving style and dynamic weight storage submodule. Otherwise, the weight self-learning submodule does not adjust the weights of the PV-reward function and directly transfers the corresponding weights and the corresponding driving style quantitative data to the driving style and dynamic weight storage submodule. The lane-changing performance baseline evaluation module operates as follows: First, the kinematic information of the target lane's following vehicle, the ego vehicle's kinematic information, the kinematic information of surrounding vehicles, and the driving style quantification data from the driving style quantification module extracted from the dataset are used as the current state. After data preprocessing, the data is passed to the trained data-driven lane-changing decision submodule. The trained data-driven lane-changing decision submodule outputs the current SV action, i.e., lane change or lane keeping, based on the current state. The action is then passed to the decision evaluation submodule. The decision evaluation submodule compares the SV's decision of whether to change lanes or keep lanes, as input from the decision judgment submodule of the lane-changing decision module, with the output of the data-driven lane-changing decision submodule to evaluate which of the two is better. The evaluation result is then output to the baseline self-evaluator and the weight self-learning submodule of the lane-changing decision module. The weight self-learning submodule dynamically adjusts the weights of the PV reward function based on the evaluation result of the decision evaluation submodule. The baseline self-evaluator determines whether the current weight self-evolution process can be terminated based on the evaluation result of the decision evaluation submodule.

2. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 1, characterized in that: The data feature extraction function of the kinematic feature extraction submodule is to obtain the relative speed, relative acceleration, and jerk information of the vehicle behind the target lane relative to the vehicle within time T, calculate the average relative speed, average relative acceleration, and average jerk, and transmit the calculated average relative speed, average relative acceleration, and average jerk information to the driving style assessment submodule.

3. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 1, characterized in that: The training process of the Gaussian mixture model is as follows: Step 1: Preprocessing and feature extraction: Obtain the relative speed, relative acceleration, and jerk information of the target lane following vehicle with N lane-changing behaviors from the dataset within T time before the ego vehicle changes lanes, and calculate the average relative speed of the target lane following vehicle in the i-th training data. Average relative acceleration and average jerk Recorded as All N training data that meet the conditions are expressed as: P = [P1, P2, ..., P n ]; Step 2: Parameter training of Gaussian mixture model: All N qualified training data are passed into the Gaussian mixture model. The number of sub-Gaussian models K in the Gaussian mixture model is set to 3. The maximum expectation algorithm is used to solve the parameters of the Gaussian mixture model. It is mainly divided into two steps: The first step, E-step, calculates the Q function according to the following formula, which represents the probability that the i-th data belongs to the k-th sub-Gaussian distribution: Among them, φ k represents the probability of the kth sub-Gaussian distribution, x (i) represents the i-th data, μ k represents the mean of the kth sub-Gaussian distribution, σ k Represents the standard deviation of the k-th sub-Gaussian distribution, and φ is randomly given in the first round of iteration k 、μ k , σ k The value of The second step, M-step, updates the model parameters according to the following formula: Repeat the E-step and M-step iterations several times until the model converges; After the training is completed, the Gaussian mixture model evaluates the probability x that the following vehicle's driving style is cautious by inputting the following vehicle's average relative speed, average relative acceleration, and average jerk information within the time T before the vehicle changes lanes. d1 , the general probability x d2 , the probability of radical type x d3 , and output the above probability as the output of the driving style evaluation submodule.

4. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 1, characterized in that: At the current time step t, the lane change benefit function submodule calculates the SV benefit function result and PV benefit function result of the current time step t based on the kinematic information of the target lane vehicle, the ego vehicle kinematic information, the surrounding vehicle kinematic information, and the current driving style quantification data obtained from the driving style quantification module. The established ego vehicle SV benefit function and PV benefit function are as follows: SV return function S priceSV for: In the above formula, K SV1 , K SV2 , K SV3 , K p1 , K p2 Both are proportional coefficients, R lane is the lane-changing benefit of SV when making lane-changing decisions; V S (v SV ,v FV1 ) function represents the desired benefit of the ego vehicle SV to ensure its own driving speed by changing lanes, which is expressed as: Among them, v SV is the velocity of the ego vehicle SV at the current time step t, v FV1 is the speed of the preceding vehicle FV1 in the lane where the ego vehicle was before changing lanes at the current time step t; V FV (v FV1 ,v FV2 ) function represents the benefit of the vehicle’s lane-changing action in improving its subsequent following speed, which is expressed as: Among them, v FV1 is the velocity of FV1 at the current time step t, v FV2 is the speed of the front vehicle FV2 in the target lane at the current time step t; S PV (S SV ,S PV ) function represents the safety benefit based on the longitudinal distance between the ego vehicle SV and the following vehicle PV, which is expressed as: Among them, S SV is the longitudinal position of SV at the current time step t, S PV is the longitudinal position of PV at the current time step t; S PV (S SV ,S PV ) is a piecewise function. When the longitudinal distance S between SV and PV is SV -S PV Exceeding the safety threshold S s Provides a constant positive return R sp ; When the longitudinal distance S between SV and PV SV -S PV At the safety threshold S s With the danger threshold S d When the spacing is between, it provides a positive benefit that gradually decreases as the spacing decreases. Until the risk threshold is reached, the profit drops to 0; when the vertical distance S between SV and PV SV -S PV Below the danger threshold S d When a large negative constant return R is provided spd ;S p1 (S PV1 ,S SV ) function represents the SV behavior decision constraint function based on the longitudinal distance and longitudinal speed between SV and FV1, which is expressed as: Among them, S FV1 is the longitudinal position of FV1 at the current time step t. When the longitudinal arrival time of SV and FV1 is When it is less than 3.5 seconds, S p1 (S FV1 ,S SV ) provides a large negative return R spn1 ; When the longitudinal arrival time of SV and FV1 When it is greater than or equal to 3.5 seconds, S p1 (S FV1 ,S SV ) provides a small positive return R sp1 ;S p2 (S FV2 ,S SV ) function represents the SV behavior decision constraint function based on the longitudinal distance and longitudinal speed between SV and FV2, which is expressed as: Among them, S FV2 is the longitudinal position of FV2 at the current time step t. When the longitudinal arrival time of SV and FV2 is Less than the minimum lane-changing safety following time difference t sp2 =3.5+t lanec When S p2 (S FV2 ,S SV ) provides a large negative return R spn2 ; When the longitudinal arrival time of SV and FV2 Greater than or equal to the minimum lane-changing safety following time difference t sp2 hour, Provides a small positive return R sp2 , where t lanec is the average lane changing time of the driver, t lanec The average lane changing time of drivers is obtained by counting the public data set; When the SV does not change lanes and stays in the original lane, it obtains a constant small negative benefit R stay ; PV return function S pricePV for: S pricePV =ω1·V PV (a PV ,v PV )+ω2·S Space (S PV ,S SV ) In the above formula, ω1 and ω2 represent the weights of the PV profit function and are dynamically variable positive weight coefficients. PV (a PV ,v PV ) function characterizes the required return on PV speed, which is expressed as: Among them, v PV is the velocity of PV at the current time step t, a PV is the acceleration value of PV at the current time step t. When the action of PV is to avoid deceleration, a PV is a negative constant. When the PV action is not to avoid, a PV =0, τ0 is a positive constant; S Space (S PV ,S SV ) function represents the required benefit of PV, which is expressed as: Among them, h0 is a positive constant, S SV is the longitudinal position of SV at the current time step t, S PV is the longitudinal position of PV at the current time step t.

5. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 1, characterized in that: The decision-making process of the decision judgment submodule is as follows: the decision judgment submodule obtains the SV benefit function result and PV benefit function result of the current time step t from the lane change benefit function submodule, and first determines the lane change benefit R of the SV at this time. lane Is it greater than the lane keeping benefit R of staying in the original lane? stay , if the lane-changing benefit of SV is R lane No greater than the lane keeping benefit R of staying in the original lane stay , the decision-making submodule determines that the lane-changing prerequisite is not met at this time, and outputs the decision that the SV should keep the lane; if the lane-changing benefit R lane Greater than the lane keeping benefit R of staying in the original lane stay , the decision judgment submodule determines that the prerequisite for lane changing is met at this time; then the PV avoidance deceleration benefit and the PV non-avoidance benefit are calculated, and the benefit values ​​of the two are compared. It is considered that PV will take the action with the larger benefit value: if the PV avoidance deceleration benefit is greater than or equal to the non-avoidance benefit, then the PV is judged to take the avoidance deceleration action; if the PV avoidance deceleration benefit is less than the non-avoidance benefit, then the PV is judged to take the non-avoidance action. Based on this, a decision is made on whether the ego vehicle SV should change lanes or continue lane keeping at this time: if the PV action is judged to be avoidance deceleration, the output decision of the decision judgment submodule at this time is lane changing; if the PV action is judged to be non-avoidance, then the output decision of the decision judgment submodule at this time is lane keeping. Then the decision judgment submodule transmits the decision to the decision evaluation submodule of the lane changing performance baseline evaluation module, and transmits the weights of the PV benefit function to the weight self-learning submodule.

6. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 1, characterized in that: The data preprocessing process before being passed to the trained data-driven lane change decision submodule is as follows: The target lane rear vehicle kinematic information, the vehicle kinematic information, the surrounding vehicle kinematic information, and the driving style quantification data X obtained from the driving style quantification module are combined into a data set. d Composition state data set S D : Among them, i represents that the state is the i-th time series sequence extracted from the data set, 0≤i≤N E ; For the i-th time sequence state sequence S D(i) , expressed as: Where t represents the current state data, which is the state data of the t-th sampling point, 0≤t≤T E , the i-th time series S D(i,t) The state data at the tth sampling point is expressed as: S D(i,t) =[And PV(i,t) ,AND SV(i,t) ,AND FV1(i,t) ,AND FV2(i,t) ,X d(i) ] Among them, E PV(i,t) represents the kinematic information of the vehicle PV behind the target lane at the tth sampling point, E SV(i,t) represents the kinematic information of the ego vehicle SV at the tth sampling point, E FV1(i,t) represents the kinematic information of the preceding vehicle FV1 in the lane before the ego vehicle changes lanes at the tth sampling point, E FV2(i,t) represents the kinematic information of the front vehicle FV2 in the target lane at the tth sampling point, X d(i) represents the driving style quantitative data of the PV of the i-th time series.

7. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 1, characterized in that: The training process of the data-driven lane-changing decision submodule is as follows: First, from the state data set S D Take out any time sequence S D(i) , and then the state data of each sampling point of the time series state sequence is sequentially transmitted to the data-driven lane change decision submodule in time sequence; at the tth time step, S D(i,t) The deep Q network in the data-driven lane-changing decision submodule calculates the value of the current action state. And make an action a t , where action a t is one of two discrete actions: lane keeping or lane changing. is the model parameter of the deep Q network, and then comes the t+1th time step, S D(i,t+1) The deep Q network in the data-driven lane-changing decision submodule obtains the reward function R S (t) Define the true reward r t , based on maximizing the action state value at this time The principle of estimating action a t+1 =argmaxQ(s t+1 ,a,ω), calculate the timing target Where 0<γ<1, is the minimum loss Loss: Perform gradient descent to update model parameters: Where η is a positive constant; Repeat the above steps until the model parameters converge, and the data-driven lane-changing decision submodule is trained. After the training is completed, the S D(i,t) The data-driven lane-changing decision submodule is input, and the data-driven lane-changing decision submodule outputs the decision at time step t.

8. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 7, characterized in that: The reward function R of the deep Q network is S (t) is defined as follows: Among them, μ1, μ2, μ3 are constant coefficients, v SV (t) indicates the time from E SV(i,t) The obtained SV longitudinal velocity information, v FV1 (t) indicates the time from E FV1(i,t) The longitudinal speed information of the vehicle FV1 in the lane before the vehicle changes lanes is obtained, v FV2 (t) indicates the time from E FV2(i,t) The longitudinal speed information of the vehicle FV2 in the target lane is obtained, S SV (t) indicates the time from E SV(i,t) The longitudinal position information of the SV obtained, S PV (t) indicates the time from E PV(i,t) The longitudinal position information of PV obtained, S d is the danger threshold.

9. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 1, characterized in that: The evaluation process of the decision evaluation submodule is as follows: The decision evaluation submodule receives the decision action A1 from the decision judgment submodule of the lane change decision module and the decision action A2 from the data-driven lane change decision submodule. Based on the output values ​​of these two actions, it determines the quality of the action according to the following judgment rules and takes action: (1) If A1 = A2 = lane keeping, then at this moment, since lane changing has not occurred, the decision evaluation submodule only outputs decision-consistent information to the baseline self-evaluator; (2) If A1 = lane keeping and A2 = lane changing, then the decision evaluation submodule uses the cubic polynomial method to keep the speed at the start and completion of the lane change consistent, and the lane change time is t lanec Plan a lane change route and attempt a lane change; during the lane change attempt, the trajectories of surrounding vehicles and PV are updated based on the kinematic information obtained from the dataset, and the SV's trajectory is updated based on the planned lane change route: If the longitudinal distance between SV and PV falls below the safety threshold S during the lane change attempt, d If the lane-changing decision module is superior to lane keeping, the decision evaluation submodule only outputs the information that the lane-changing decision module is superior to the baseline self-evaluator. If the longitudinal distance between SV and PV is always higher than the safety threshold S during the lane change attempt d , then it is considered that lane changing is better than lane keeping at this moment. Therefore, the decision evaluation submodule outputs the information that lane changing is currently possible to the weight self-learning submodule of the lane change decision module, and outputs the information that the decision is inconsistent to the baseline self-evaluator; (3) If A1 = lane change and A2 = lane change, then at this moment, since the actions are consistent, the decision evaluation submodule only outputs the decision consistency information to the baseline self-evaluator; (4) If A1 = lane change and A2 = lane keeping, then the decision evaluation submodule uses a cubic polynomial to plan the lane change route and attempts a lane change. During the lane change attempt, the trajectories of the surrounding vehicles and the PV are updated based on the kinematic information obtained from the dataset, and the SV's trajectory is updated based on the planned lane change route. If the longitudinal distance between the SV and the PV falls below the safety threshold S during the lane change attempt, d If the lane change is not performed correctly, it is considered that lane changing is inferior to lane keeping. Therefore, the decision evaluation submodule outputs information about the risk of lane changing to the weight self-learning submodule of the lane change decision module, and outputs information about inconsistent decisions to the baseline self-evaluator. If the longitudinal distance between SV and PV is always higher than the safety threshold S during the lane change attempt d , then it is considered that lane changing is better than lane keeping at this moment, so the decision evaluation submodule only outputs the information that the lane changing decision module is better to the baseline self-evaluator; The baseline self-assessor operates as follows: The baseline self-evaluator receives the information output by the decision evaluation submodule and counts the number of information h1 that is better than the lane change decision module, the number of information h2 that is inconsistent with the decision, and the number of information h3 that is consistent with the decision every M data. If and It is considered that the decision of the decision judgment submodule is significantly better than the decision of the data-driven lane change decision submodule at this time, and the weight self-evolution process is terminated; otherwise, the weight self-evolution process continues; where β0 and β1 are positive constants greater than zero.

10. The weighted self-evolving intelligent vehicle personalized lane-changing decision system according to claim 1, characterized in that: The operation process of the weight self-learning submodule is as follows: The weight self-learning submodule receives the weights of the PV benefit function of the decision judgment submodule and the information output by the decision evaluation submodule: (1) When receiving the information from the decision evaluation submodule that the current lane change is risky, it is considered that the weight of the speed benefit term in the PV benefit function is underestimated. Therefore, the weight self-learning submodule adjusts the PV benefit function S according to the following formula: pricePV The weights ω1 and ω2 are the new weights ω'1 and ω'2: 1-ω'1→ω'2, where α is a positive constant coefficient; (2) When receiving the information that the lane can be changed from the decision evaluation submodule, it is considered that the weight of the distance benefit term in the PV benefit function is underestimated. Therefore, the weight self-learning submodule adjusts the PV benefit function S according to the following formula pricePV The weights ω1 and ω2 are the new weights ω'1 and ω'2: 1-ω'2→ω'1, where α is a positive constant coefficient; (3) When no information is received from the decision evaluation submodule, the weight self-learning submodule does not adjust the weights of the PV profit function; Subsequently, the weight self-learning submodule transfers the latest PV benefit function weights and their corresponding driving style quantitative data to the driving style and dynamic weight storage submodule for storage; The operation process of the driving style and dynamic weight storage submodule is as follows: First, the driving style and dynamic weight storage submodule receives the newly passed PV reward function weights and their corresponding driving style quantitative data from the weight self-learning submodule. Then, the driving style and dynamic weight storage submodule queries the current storage for driving style quantitative data records that match the newly passed driving style quantitative data and performs the following operations: (1) If there is a driving style quantification data record that is consistent with the newly input driving style quantification data, the driving style and dynamic weight storage submodule updates the original record, and the weights of the newly input PV benefit function are used to overwrite the weight records of the original PV benefit function; (2) If there is no driving style quantitative data record that is consistent with the newly-input driving style quantitative data, the driving style and dynamic weight storage submodule stores the weights of the newly-input PV benefit function and its corresponding driving style quantitative data as a record in the storage.

Images