A method for intelligent vehicle trajectory planning based on driving intention prediction model

Abstract

The present invention discloses a method for planning the driving trajectory of an intelligent vehicle based on a driving intention prediction model. By combining the driving emotion transfer probability distribution and the driving intention probability distribution, it realizes the accurate prediction of the driver's complex intentions and emotional changes, and integrates them into the trajectory planning process. By statistically analyzing the driver's historical driving data, the driving emotion transfer probability distribution and the driving intention probability distribution are determined, and the driver's future driving intention and emotional state are accurately predicted. The driver's emotional state and intention changes are comprehensively considered to generate a trajectory that meets the driver's expectations, thereby improving the driving experience and safety. A multi-objective optimization function is designed to incorporate safety, comfort and intention compliance into a unified framework, and the optimal trajectory is generated in real time through an optimization algorithm. Environmental perception information and uncertainty processing methods are introduced to improve the robustness and reliability of trajectory planning in complex dynamic environments.

Description

Technical Field

[0001] The present invention relates to a method for planning a driving trajectory of an intelligent vehicle based on a driving intention prediction model. Background Art

[0002] With the continuous development of automobile technology, autonomous driving technology is gradually becoming an important direction for future intelligent transportation. However, current autonomous driving systems still have many shortcomings in aspects such as trajectory planning combined with driving intention prediction. Autonomous driving systems need to accurately understand and predict the driver's intentions to ensure safety and comfort. The prediction of driving intention is highly uncertain and is affected by many factors, including the driver's emotions, environmental changes, and traffic conditions. Existing driving intention prediction methods mostly rely on the vehicle's motion state and simple environmental perception information, which makes it difficult to fully capture the driver's complex intention changes. At the same time, the driver's emotional state (such as tension, happiness, fatigue, etc.) has a significant impact on their driving behavior. If the autonomous driving system cannot effectively consider the driver's emotional changes, it may lead to system decisions that are inconsistent with the driver's expectations, affecting the driving experience and safety. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for intelligent vehicle driving trajectory planning based on a driving intention prediction model to solve the problems faced in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A method for intelligent vehicle driving trajectory planning based on a driving intention prediction model includes the following steps:

[0006] S101, obtaining current vehicle status, environmental information, and driver's driving emotion;

[0007] S201. Predicting future emotional states based on driving emotion transfer probability distribution;

[0008] S301, predicting possible future driving intentions based on a driving intention probability distribution;

[0009] S401, defining a typical target trajectory cluster based on driving intention, generating candidate trajectories, and calculating trajectory deviations;

[0010] S501. Design a multi-objective optimization function based on driving safety, driving comfort, and driving intention compliance;

[0011] S601, determining trajectory planning constraints;

[0012] S701, minimizing the objective function to find the optimal trajectory while satisfying the constraints;

[0013] S801: Tracking the obtained overtaking trajectory using an MPC controller;

[0014] The vehicle state information in step S101 includes the position, speed, and acceleration of the current vehicle and surrounding vehicles; the environmental information includes the current road scene obtained through the vehicle-mounted environmental perception system and the high-definition map of the unmanned intelligent vehicle; the driver's driving emotions include the following 8 types: joy Emo1, relaxation Emo2, provocation Emo3, helplessness Emo4, anxiety Emo5, fear Emo6, panic Emo7, and anger Emo8; the driver's lane change intention includes the following 7 types: maintain the original speed and go straight Int1, accelerate and go straight Int2, accelerate and go straight Int3, maintain the original speed and change lanes Int4, accelerate and change lanes Int5, decelerate and change lanes Int6, and stop Int7;

[0015] Steps S201 and S301 are the calculation process of determining the driving emotion transfer probability and driving intention probability distribution through the hidden Markov model. The specific values ​​are determined through experiments. The basic parameters in the hidden Markov model are defined as follows:

[0016] Emo={Emo1, Emo2,…, Emo8}, where Emo is the set of the driver’s driving emotional states;

[0017] Int={Int1, Int2,…, Int7}, where Int is the driver’s driving intention state set;

[0018] At time t, the driver's driving emotional state is x t , the driving intention state at time t is y t ;

[0019] η i (s k )=P(y k =s k |x k =Emo i ),i=1,2,…,8,k=1,2,…,t,…,L, where η i (s k ) is the driver’s emotional state at time k, Emo i Driving intention k =s k The probability of s k ∈{Int1,Int2,…,Int7};

[0020] where ε i The driving emotion Emo of the driver during vehicle operation i The probability of occurrence, M iis the driving emotion state sequence x1,x2,…,x t There is M i Data for Emo i ;

[0021] where ε ij is the driving emotion state sequence x1,x2,…,x t In the process of vehicle operation, the driving emotional state is determined by Emo i Transformed into Emo j The probability of M ij For M i There are M ij The data is transferred to Emo at the next moment j ;

[0022] where η ijm For the driving intention state sequence y1,y2,...,y t In the process of vehicle operation, the driver's driving emotion is expressed by Emo i Transformed into Emo j When the driver's driving intention Int m The probability of generation, N ijm Driving Emotion for Drivers by Emo i Transformed into Emo j The driver's driving intention is Int m The number of

[0023] where η ijmn For the driving intention state sequence y1,y2,...,y t In the process of vehicle operation, the driver's driving emotion is expressed by Emo i Transformed into Emo j When the driver's driving intention is represented by Int m Convert to Int n The probability, N ijmn For N ijm There are N data ijmn The driver's driving intention at the next moment is represented by Int m Convert to Int n The number of

[0024] α t (i) = P(y k =s k |x t =Emo i ),k=1,2,…,t, where α t(i) The driving emotional state at time t is Emo i The driver’s driving intention y at time k (k=1,2,…,t) k =s k The probability of , and the initial value α1(i)=ε i η i (s1), recursive relation

[0025] β t (i) = P(y k =s k ∣x t =Emo i ),k=t+1,t+2,…,L, where β t (i) The driving emotional state at time t is Emo i The driver's driving intention y at time k (k=t+1,t+2,…,L) k =s k The probability of , and has a terminal value β L (i)=1, recursive relationship

[0026]

[0027] 1≤t≤L-1, where σ t (i,j) is the number of cells in the known driving intention sequence s1,s2,...,s t , the driving emotional state at time t+1 is determined by Emo i Transformed into Emo j probability;

[0028] where ξ t (i) is the driving intention sequence s1,s2,...,s t , the driving emotional state at time t is Emo i The probability of

[0029] The driving emotion transfer probability distribution in step S201 is determined by the following formula:

[0030]

[0031] The driving intention probability distribution in step S301 is determined by the following formula:

[0032]

[0033] The typical target trajectory cluster defined by the driving intention in step S401 is determined by the following formula:

[0034] Use the car's linear two-degree-of-freedom model to describe the vehicle's motion state:

[0035]

[0036] Among them, (x V ,y V ) is the vehicle's position, v is the driving speed, a is the driving acceleration, θ is the vehicle's heading angle, and ω is the steering angular velocity;

[0037] Maintaining the original speed and driving straight ahead, the typical target trajectory cluster of Int1 is determined by the following formula:

[0038] x V,t+1 =x V,t +v t cos(θ t )Δt

[0039] y V,t+1 =y V,t +v t sin(θ t )Δt

[0040] v t+1 =v t +a t Δt

[0041] θ t+1 =θ t +ω t Δt

[0042] Among them, x V,t is the lateral position of the vehicle at time t, y V,t is the longitudinal position of the vehicle at time t, Δt is the time interval, v t is the driving speed at time t, a t is the driving acceleration at time t, and a t =0,ω t is the turning angular velocity at time t, and ω t =0;

[0043] The typical target trajectory cluster of accelerating straight Int2 is determined by the following formula:

[0044] x V,t+1 =x V,t +v t cos(θ t )Δt

[0045] y V,t+1 =y V,t +v t sin(θ t )Δt

[0046] v t+1 =v t +a t Δt

[0047] θ t+1 =θ t +ω t Δt

[0048] Among them, a t =a0, a0 is the acceleration during acceleration, ω t =0;

[0049] The typical target trajectory cluster of deceleration and straight driving Int3 is determined by the following formula:

[0050] x V,t+1 =x V,t +v t cos(θ t )Δt

[0051] y V,t+1 =y V,t +v t sin(θ t )Δt

[0052] v t+1 =v t +a t Δt

[0053] θ t+1 =θ t +ω t Δt

[0054] Among them, a t =-a0,ω t =0;

[0055] The typical target trajectory cluster of Int4 for lane change while maintaining the original speed is determined by the following formula:

[0056] x V,t+1 =x V,t +v t cos(θ t )Δt

[0057] y V,t+1 =y V,t +v t sin(θ t )Δt

[0058] v t+1 =v t +a t Δt

[0059] θ t+1 =θ t +ωt Δt

[0060] Among them, a t =0, ω change is the steering angular velocity during lane change; the typical target trajectory cluster of accelerated lane change Int5 is determined by the following formula:

[0061] x V,t+1 =x V,t +v t cos(θ t )Δt

[0062] y V,t+1 =y V,t +v t sin(θ t )Δt

[0063] v t+1 =v t +a t Δt

[0064] θ t+1 =θ t +ω t Δt

[0065] Among them, a t =a0, The typical target trajectory cluster of deceleration lane change Int6 is determined by the following formula:

[0066] x V,t+1 =x V,t +v t cos(θ t )Δt

[0067] y V,t+1 =y V,t +v t sin(θ t )Δt

[0068] v t+1 =v t +a t Δt

[0069] θ t+1 =θ t +ω t Δt

[0070] Among them, a t =-a0,

[0071] The typical target trajectory cluster of parking Int7 is determined by the following formula:

[0072] x V,t+1=x V,t

[0073] y V,t+1 =y V,t

[0074] v t+1 =0

[0075] θ t+1 =θ t

[0076] The target trajectories generated above are grouped into {τ1,τ2,...,τ7}, where each trajectory τ i Corresponding driving intention Int i .

[0077] The trajectory deviation calculated in step S401 is determined by the following formula:

[0078]

[0079] Among them, d τ (t,i) is the trajectory deviation, (x V,t ,y V,t ) is the current vehicle position, is the generated candidate trajectory τ.

[0080] The driving safety index in the multi-objective optimization function of step S501 is determined by the following formula:

[0081]

[0082] Among them, J safety is the driving safety index, d obs (t) is the distance between the point on the planned trajectory and the obstacle, d lane (t) The distance from the point on the planned trajectory to the centerline of the lane, w obs , w lane is the weight coefficient.

[0083] The driving comfort index in the multi-objective optimization function of step S501 is determined by the following formula:

[0084]

[0085] Among them, J comfort is the comfort index, a x (t) is the lateral acceleration of the car at time t, a x,desired is the expected lateral acceleration of the vehicle, a y (t) is the longitudinal acceleration of the vehicle at time t, a y,desired is the expected longitudinal acceleration of the car, ω(t) is the steering angular velocity of the car at time t, ω desiredis the expected vehicle steering acceleration, w ω is the weight coefficient.

[0086] The driving intention compliance in the multi-objective optimization function of step S501 is determined by the following formula:

[0087]

[0088] Among them, J intention is the driving intention compliance, w τ is the weight coefficient.

[0089] The multi-objective optimization function of step S501 is determined by the following formula:

[0090] J=q s ·J safety +q c ·J comfort +q i ·J intention

[0091] Among them, q s ,q c ,q i is the weight coefficient.

[0092] The trajectory planning constraints in step S601 are composed of the following constraints:

[0093] A. Vehicle dynamics constraints, including speed constraints, acceleration constraints, and steering angular velocity constraints;

[0094] B. Environmental constraints, including lane constraints, obstacle avoidance constraints, and collision constraints;

[0095] C. Comfort constraints, including acceleration rate constraints;

[0096] D. Lane change intention constraints, including driver driving intention constraints and lane change completion time constraints.

[0097] Compared with the prior art, the present invention has the following beneficial effects:

[0098] 1. This invention constructs a driving emotion transition probability matrix and a driving intention probability distribution matrix, and performs statistical analysis based on a large amount of actual driving data to more accurately predict a driver's future driving intention and emotional state. This precise prediction improves the reliability of trajectory planning, enabling the autonomous driving system's decisions to better meet the driver's actual needs.

[0099] 2. This invention comprehensively considers the driver's emotional state and changing intentions during trajectory planning. The resulting trajectory is not only safe and stable, but also more aligned with the driver's expectations. By incorporating the driver's emotions and intentions into the optimization objective function, the driving experience and system humanization are significantly enhanced, reducing conflicts between the driver and the autonomous driving system.

[0100] 3. This invention incorporates environmental perception information and uncertainty processing methods, resulting in trajectories with enhanced robustness and reliability in complex dynamic environments. This multi-objective optimization function not only considers safety, comfort, and intent compliance, but also utilizes real-time optimization algorithms to enhance the system's adaptability to perception errors and environmental changes, ensuring safe driving in a variety of complex traffic scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] The present invention will be further described below in conjunction with the accompanying drawings:

[0102] Figure 1 This invention proposes a method for intelligent vehicle driving trajectory planning based on a driving intention prediction model. DETAILED DESCRIPTION

[0103] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0104] like Figure 1 As shown, the present invention provides the following technical solution: a method for planning the driving trajectory of an intelligent vehicle based on a driving intention prediction model, comprising the following steps:

[0105] S101, obtaining current vehicle status, environmental information, and driver's driving emotion;

[0106] S201. Predicting future emotional states based on driving emotion transfer probability distribution;

[0107] S301, predicting possible future driving intentions based on a driving intention probability distribution;

[0108] S401, defining a typical target trajectory cluster based on driving intention, generating candidate trajectories, and calculating trajectory deviations;

[0109] S501. Design a multi-objective optimization function based on driving safety, driving comfort, and driving intention compliance;

[0110] S601, determining trajectory planning constraints;

[0111] S701, minimizing the objective function to find the optimal trajectory while satisfying the constraints;

[0112] S801: Tracking the obtained overtaking trajectory using an MPC controller;

[0113] The vehicle state information in step S101 includes the position, speed, and acceleration of the current vehicle and surrounding vehicles; the environmental information includes the current road scene obtained through the vehicle-mounted environmental perception system and the high-definition map of the unmanned intelligent vehicle; the driver's driving emotions include the following 8 types: joy Emo1, relaxation Emo2, provocation Emo3, helplessness Emo4, anxiety Emo5, fear Emo6, panic Emo7, and anger Emo8; the driver's lane change intention includes the following 7 types: maintain the original speed and go straight Int1, accelerate and go straight Int2, accelerate and go straight Int3, maintain the original speed and change lanes Int4, accelerate and change lanes Int5, decelerate and change lanes Int6, and stop Int7;

[0114] Steps S201 and S301 are the calculation process of determining the driving emotion transfer probability and driving intention probability distribution through the hidden Markov model. The specific values ​​are determined through experiments. The basic parameters in the hidden Markov model are defined as follows:

[0115] Emo={Emo1, Emo2,…, Emo8}, where Emo is the set of the driver’s driving emotional states;

[0116] Int={Int1, Int2,…, Int7}, where Int is the driver’s driving intention state set;

[0117] At time t, the driver's driving emotional state is x t , the driving intention state at time t is y t ;

[0118] η i (s k )=P(y k =s k |x k =Emo i ),i=1,2,…,8,k=1,2,…,t,…,L, where η i (s k ) is the driver’s emotional state at time k, Emo i Driving intention k =s k The probability of s k ∈{Int1,Int2,…,Int7};

[0119] where ε i The driving emotion Emo of the driver during vehicle operation i The probability of occurrence, M i is the driving emotion state sequence x1,x2,…,x tThere is M i Data for Emo i ;

[0120] where ε ij is the driving emotion state sequence x1,x2,…,x t In the process of vehicle operation, the driving emotional state is determined by Emo i Transformed into Emo j The probability of M ij For M i There are M ij The data is transferred to Emo at the next moment j ;

[0121] where η ijm For the driving intention state sequence y1,y2,...,y t In the process of vehicle operation, the driver's driving emotion is expressed by Emo i Transformed into Emo j When the driver's driving intention Int m The probability of generation, N ijm Driving Emotion for Drivers by Emo i Transformed into Emo j The driver's driving intention is Int m The number of

[0122] where η ijmn For the driving intention state sequence y1,y2,...,y t In the process of vehicle operation, the driver's driving emotion is expressed by Emo i Transformed into Emo j When the driver's driving intention is represented by Int m Convert to Int n The probability, N ijmn For N ijm There are N data ijmn The driver's driving intention at the next moment is represented by Int m Convert to Int n The number of

[0123] α t (i) = P(y k =s k |x t =Emo i ),k=1,2,…,t, where α t (i) The driving emotional state at time t is Emo iThe driver’s driving intention y at time k (k=1,2,…,t) k =s k The probability of , and the initial value α1(i)=ε i η i (s1), recursive relation

[0124] β t (i) = P(y k =s k ∣x t =Emo i ),k=t+1,t+2,…,L, where β t (i) The driving emotional state at time t is Emo i The driver's driving intention y at time k (k=t+1,t+2,…,L) k =s k The probability of , and has a terminal value β L (i)=1, recursive relationship

[0125]

[0126] 1≤t≤L-1, where σ t (i,j) is the number of cells in the known driving intention sequence s1,s2,...,s t , the driving emotional state at time t+1 is determined by Emo i Transformed into Emo j probability;

[0127] where ξ t (i) is the driving intention sequence s1,s2,...,s t , the driving emotional state at time t is Emo i The probability of

[0128] The driving emotion transfer probability distribution in step S201 is determined by the following formula:

[0129]

[0130] The driving intention probability distribution in step S301 is determined by the following formula:

[0131]

[0132] The typical target trajectory cluster defined by the driving intention in step S401 is determined by the following formula:

[0133] Use the car's linear two-degree-of-freedom model to describe the vehicle's motion state:

[0134]

[0135] Among them, (x V ,y V ) is the vehicle's position, v is the driving speed, a is the driving acceleration, θ is the vehicle's heading angle, and ω is the steering angular velocity;

[0136] Maintaining the original speed and driving straight ahead, the typical target trajectory cluster of Int1 is determined by the following formula:

[0137] x V,t+1 =x V,t +v t cos(θ t )Δt

[0138] y V,t+1 =y V,t +v t sin(θ t )Δt

[0139] v t+1 =v t +a t Δt

[0140] θ t+1 =θ t +ω t Δt

[0141] Among them, x V,t is the lateral position of the vehicle at time t, y V,t is the longitudinal position of the vehicle at time t, Δt is the time interval, v t is the driving speed at time t, a t is the driving acceleration at time t, and a t =0,ω t is the turning angular velocity at time t, and ω t =0;

[0142] The typical target trajectory cluster of accelerating straight Int2 is determined by the following formula:

[0143] x V,t+1 =x V,t +v t cos(θ t )Δt

[0144] y V,t+1 =y V,t +v t sin(θ t )Δt

[0145] v t+1 =v t +at Δt

[0146] θ t+1 =θ t +ω t Δt

[0147] Among them, a t =a0, a0 is the acceleration during acceleration, ω t =0;

[0148] The typical target trajectory cluster of deceleration and straight driving Int3 is determined by the following formula:

[0149] x V,t+1 =x V,t +v t cos(θ t )Δt

[0150] y V,t+1 =y V,t +v t sin(θ t )Δt

[0151] v t+1 =v t +a t Δt

[0152] θ t+1 =θ t +ω t Δt

[0153] Among them, a t =-a0,ω t =0;

[0154] The typical target trajectory cluster of Int4 for lane change while maintaining the original speed is determined by the following formula:

[0155] x V,t+1 =x V,t +v t cos(θ t )Δt

[0156] y V,t+1 =y V,t +v t sin(θ t )Δt

[0157] v t+1 =v t +a t Δt

[0158] θ t+1 =θ t +ω t Δt

[0159] Among them, a t =0, ω change is the steering angular velocity during lane change;

[0160] The typical target trajectory cluster of accelerated lane change Int5 is determined by the following formula:

[0161] x V,t+1 =x V,t +v t cos(θ t )Δt

[0162] y V,t+1 =y V,t +v t sin(θ t )Δt

[0163] v t+1 =v t +a t Δt

[0164] θ t+1 =θ t +ω t Δt

[0165] Among them, a t =a0,

[0166] The typical target trajectory cluster of deceleration lane change Int6 is determined by the following formula:

[0167] x V,t+1 =x V,t +v t cos(θ t )Δt

[0168] y V,t+1 =y V,t +v t sin(θ t )Δt

[0169] v t+1 =v t +a t Δt

[0170] θ t+1 =θ t +ω t Δt

[0171] Among them, a t =-a0,

[0172] The typical target trajectory cluster of parking Int7 is determined by the following formula:

[0173] x V,t+1 =x V,t

[0174] y V,t+1 =y V,t

[0175] v t+1 =0

[0176] θ t+1 =θ t

[0177] The target trajectories generated above are grouped into {τ1,τ2,...,τ7}, where each trajectory τ i Corresponding driving intention Int i .

[0178] The trajectory deviation calculated in step S401 is determined by the following formula:

[0179]

[0180] Among them, d τ (t,i) is the trajectory deviation, (x V,t ,y V,t ) is the current vehicle position, is the generated candidate trajectory τ.

[0181] The driving safety index in the multi-objective optimization function of step S501 is determined by the following formula:

[0182]

[0183] Among them, J safety is the driving safety index, d obs (t) is the distance between the point on the planned trajectory and the obstacle, d lane (t) The distance from the point on the planned trajectory to the centerline of the lane, w obs , w lane is the weight coefficient.

[0184] The driving comfort index in the multi-objective optimization function of step S501 is determined by the following formula:

[0185]

[0186] Among them, J comfort is the comfort index, a x (t) is the lateral acceleration of the car at time t, a x,desired is the expected lateral acceleration of the vehicle, a y (t) is the longitudinal acceleration of the vehicle at time t, a y,desiredis the expected longitudinal acceleration of the car, ω(t) is the steering angular velocity of the car at time t, ω desired is the expected vehicle steering acceleration, w ω is the weight coefficient.

[0187] The driving intention compliance in the multi-objective optimization function of step S501 is determined by the following formula:

[0188]

[0189] Among them, J intention is the driving intention compliance, w τ is the weight coefficient.

[0190] The multi-objective optimization function of step S501 is determined by the following formula:

[0191] J=q s ·J safety +q c ·J comfort +q i ·J intention

[0192] Among them, q s ,q c ,q i is the weight coefficient.

[0193] The trajectory planning constraints in step S601 are composed of the following constraints:

[0194] A. Vehicle dynamics constraints, including speed constraints, acceleration constraints, and steering angular velocity constraints;

[0195] B. Environmental constraints, including lane constraints, obstacle avoidance constraints, and collision constraints;

[0196] C. Comfort constraints, including acceleration rate constraints;

[0197] D. Lane change intention constraints, including driver driving intention constraints and lane change completion time constraints.

Claims

1. A method for intelligent vehicle driving trajectory planning based on a driving intention prediction model, characterized in that: The following steps are involved: S101, obtaining current vehicle status information, environmental information, and driver's driving emotion; S201. Predicting future emotional states based on driving emotion transfer probability distribution; S301, predicting possible future driving intentions based on a driving intention probability distribution; S401, defining a typical target trajectory cluster based on driving intention, generating candidate trajectories, and calculating trajectory deviations; S501. Design a multi-objective optimization function based on driving safety, driving comfort, and driving intention compliance; S601, determining trajectory planning constraints; S701, minimizing the multi-objective optimization function to obtain the optimal trajectory while satisfying the constraints; S801: Tracking the obtained overtaking trajectory using an MPC controller; The vehicle state information in step S101 includes the position, speed, and acceleration of the current vehicle and surrounding vehicles; the environmental information includes the current road scene obtained through the vehicle's environmental perception system and the high-definition map of the unmanned intelligent vehicle; the driver's driving emotions include the following 8 types: joy Emo1, relaxed Emo2, provocative Emo3, helpless Emo4, anxious Emo5, fear Emo6, panic Emo7, and angry Emo8; The driver's lane change intention includes the following seven types: maintain the original speed and go straight Int1, accelerate and go straight Int2, decelerate and go straight Int3, maintain the original speed and change lanes Int4, accelerate and change lanes Int5, decelerate and change lanes Int6, and stop Int7; Steps S201 and S301 are the calculation process of determining the driving emotion transfer probability and driving intention probability distribution through the hidden Markov model. The specific values ​​are determined through experiments. The basic parameters in the hidden Markov model are defined as follows: , where Emo is the set of the driver’s driving emotional state; , where Int is the driver’s driving intention state set; At time t, the driver’s driving emotional state is , the driving intention state at time t is ; ,in For the driver The driving emotional state at all times is Driving intention The probability of ; ,in The driver's driving emotion during vehicle operation The probability of occurrence, Sequence of emotional states in driving There are The data is ; ,in Sequence of emotional states in driving In the process of vehicle operation, the driving emotional state is determined by Transformed into The probability of For Among the data The data is transferred to the ; ,in For the driving intention state sequence In the process of vehicle operation, the driver's driving emotion is determined by Transformed into When the driver's driving intention The probability of occurrence, The driver's driving emotion Transformed into The driver's driving intention is The number of ,in For the driving intention state sequence In the process of vehicle operation, the driver's driving emotion is determined by Transformed into When the driver's driving intention is Transformed into The probability of For Among the data The driver's driving intention at the next moment is determined by Transformed into The number of ,in The driving emotional state at time t is The driver is Driving intention at all times The probability of , and there is an initial value , recursive relation , 1≤j≤8, 1≤t≤L-1; ,in The driving emotional state at time t is The driver is Driving intention at all times The probability of , and has a terminal value , recursive relation ; 1≤t≤L-1, where For a known driving intention sequence , the driving emotional state at time t+1 is given by Transformed into probability; ,in For a known driving intention sequence , the driving emotional state at time t is The probability of ; The driving emotion transfer probability distribution in step S201 is determined by the following formula: ； The driving intention probability distribution in step S301 is determined by the following formula: ； The trajectory deviation calculated in step S401 is determined by the following formula: ， in, is the trajectory deviation, is the current vehicle position, is the generated candidate trajectory; The driving safety index in the multi-objective optimization function of step S501 is determined by the following formula: ， in, It is an indicator of driving safety. is the distance between the point on the planned trajectory and the obstacle, The distance between a point on the planned trajectory and the lane centerline, , is the weight coefficient; In step S501, the driving comfort index in the multi-objective optimization function is determined by the following formula: ， in, is the comfort index, is the lateral acceleration of the car at time t, is the expected lateral acceleration of the vehicle, is the longitudinal acceleration of the car at time t, is the expected longitudinal acceleration of the vehicle, is the steering angular velocity of the car at time t, is the expected vehicle steering acceleration, , , is the weight coefficient; The driving intention compliance in the multi-objective optimization function of step S501 is determined by the following formula: ， in, driving intention compliance, is the weight coefficient.

2. The method for intelligent vehicle driving trajectory planning based on a driving intention prediction model according to claim 1, characterized in that: The typical target trajectory cluster defined by the driving intention in step S401 is determined by the following formula: Use the car's linear two-degree-of-freedom model to describe the vehicle's motion state: ， in, The position of the vehicle, For driving speed, is the driving acceleration, is the vehicle heading angle, is the steering angular velocity; Maintaining the original speed and driving straight ahead, the typical target trajectory cluster of Int1 is determined by the following formula: ， in, is the lateral position of the vehicle at time t, is the longitudinal position of the vehicle at time t, is the time interval, is the driving speed at time t, is the driving acceleration at time t, and , is the turning angular velocity at time t, and ; The typical target trajectory cluster of accelerating straight Int2 is determined by the following formula: ， in, , is the acceleration during acceleration, ; The typical target trajectory cluster of deceleration and straight driving Int3 is determined by the following formula: ， in, , ; The typical target trajectory cluster of Int4 for lane change while maintaining the original speed is determined by the following formula: ， in, , , is the steering angular velocity during lane change; The typical target trajectory cluster of accelerated lane change Int5 is determined by the following formula: ， in, , ; The typical target trajectory cluster of deceleration lane change Int6 is determined by the following formula: ， in, , ; The typical target trajectory cluster of parking Int7 is determined by the following formula: ， The target trajectories generated above are combined into , where each trajectory Corresponding driving intention Int i .

3. The method for intelligent vehicle driving trajectory planning based on a driving intention prediction model according to claim 1, characterized in that: The multi-objective optimization function of step S501 is determined by the following formula: ,in, , , is the weight coefficient.

Images