Co-driving system permission allocation and planning control method considering driving intention and state

Abstract

The present application belongs to the technical field of vehicle control, and relates to a co-driving system permission allocation and planning control method considering driving intention and state, comprising: step 1. Driver intention recognition model construction, through multi-modal data acquisition, combining a CNN model and an LSTM model to extract spatial and temporal features, and fusing to recognize driving intention in real time; step 2. Trajectory planning decision design, judging lane changing feasibility according to the recognized driving intention, generating a safe trajectory cluster, and realizing trajectory planning; step 3. Real-time solution of man-machine optimal weight, realizing smooth cooperation and safe switching between the driver and the co-driving system; and step 4. Solution of motion control quantity, based on the optimal lane changing trajectory, using an adaptive model predictive control method to realize high-precision tracking control of the vehicle on the expected trajectory. The present application has the advantages that the LSTM model and the CNN model are fused for driver intention recognition, and the time features and the space features are fused to ensure high recognition accuracy.

Description

Technical Field

[0001] This invention belongs to the field of vehicle control technology, and relates to permission allocation, trajectory planning and motion control in human-machine co-driving vehicles. Specifically, it relates to a permission allocation and planning control method for a co-driving system that takes into account driving intentions and states. Background Technology

[0002] In human-machine co-driving mode, the driver and the autonomous driving system share driving tasks to varying degrees, collaboratively planning and controlling the vehicle. This mode leverages the driver's decision-making flexibility and intuitive judgment in complex situations, while also utilizing the data processing capabilities and high-precision control of the autonomous driving system, potentially achieving a comprehensive improvement in safety, comfort, and efficiency. However, in human-machine co-driving scenarios, especially lane-changing scenarios, unreasonable allocation of human and machine weights, improper lane-changing trajectory planning, and low trajectory tracking accuracy can all lead to human-machine conflicts. These conflicts not only degrade the driving experience but may also cause vehicle instability and even traffic accidents. Therefore, it is necessary to design a human-machine permission allocation and planning control method to ensure coordination of human and machine intentions during lane changes and to provide effective driving assistance based on the driver's real-time status. Summary of the Invention

[0003] In view of the above-mentioned technical problems and defects, the purpose of this invention is to provide a method for permission allocation and planning control of a co-driving system that considers driving intention and state. This method is based on the recognition results of the driver's lane-changing intention and designs lane-changing feasibility judgment conditions to carry out trajectory planning and motion control of the co-driving system. At the same time, it allocates human-machine control permissions based on the driver's real-time state, effectively realizing the coordination of human-machine driving intentions under the premise of vehicle safety, and providing effective support for the development and implementation of human-machine co-driving technology.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] A method for permission allocation and planning control in a co-driving system that considers driving intentions and states includes the following steps:

[0006] Step 1. Driver intent recognition model construction: Through multimodal data collection, spatial and temporal features are extracted by combining CNN and LSTM models, and the driving intent is recognized in real time after fusion.

[0007] Step 2. Trajectory planning decision design: Based on the identified driving intention, determine the feasibility of lane changing, generate a safe trajectory cluster, and use a multi-objective scoring method to select the optimal lane changing trajectory to achieve trajectory planning;

[0008] Step 3. Real-time solution of optimal human-machine weights: Based on the driver's intentions and steering wheel angle fluctuations, dynamically adjust the human-machine weight allocation to achieve smooth collaboration and safe switching between the driver and the co-driving system.

[0009] Step 4. Solve for motion control variables. Based on the optimal lane-changing trajectory, adopt an adaptive model predictive control method, and combine vehicle speed dynamic adjustment prediction and control time domain to achieve high-precision tracking control of the vehicle on the desired trajectory.

[0010] As a preferred embodiment of the present invention, step 1 further includes the following step:

[0011] Step 1.1. Real-time multimodal data acquisition: During vehicle operation, data is collected in real time using sensors: (1) Time series data: real-time status is obtained from the vehicle's CAN bus, including position, speed, acceleration, steering wheel angle, and light signals; (2) Spatial feature data: real-time images, lane lines, and relative positions of surrounding vehicles are obtained through environmental perception devices. The time series data is sliced ​​according to a fixed time window to form a sliding time window; at the same time, the spatial feature data is dynamically updated frame by frame to match the time series data. The environmental perception devices include cameras, radar, or lidar.

[0012] Step 1.2. Construct a CNN model to process spatial features; preprocess the input image data by scaling and normalizing to adapt to the input format of the CNN model, use convolutional layers to extract local spatial features, including lane line shape, relative position of vehicles in front or on the side, use pooling layers to reduce dimensionality, retain the preset key features, and reduce the amount of computation to meet real-time requirements, and finally obtain spatial feature vectors.

[0013] Step 1.3. Construct an LSTM model to process time series features. According to the sliding time window, input the time series data of the most recent time into the LSTM model. Use multiple LSTM modules of the LSTM model to process the time series in sequence, capture the dynamic change trend of driving behavior, and output the time series feature vector to represent the continuous dynamic features of the vehicle.

[0014] Step 1.4. Real-time feature fusion: At the end of each time window, the features extracted by the LSTM model and the CNN model are fused. The time series feature vector output by the CNN model and the spatial feature vector output by the CNN model are combined by feature concatenation. The fused feature vector is input into the fully connected layer for further processing and compression to form high-level semantic features.

[0015] Step 1.5. Classify and output the recognition results. Based on the fused high-level semantic features, predict driving intentions in real time. Use a fully connected layer to map the high-level semantic features to the classification space. Calculate the probability of each driving intention category through the Softmax activation function. The output driving intention categories include: left lane change, right lane change, and keep straight.

[0016] As a preferred embodiment of the present invention, step 2 further includes the following step:

[0017] Step 2.1. Input the intent recognition result: Input the driving intent category output by the driver intent recognition model into the trajectory planning decision module. The driving intent categories include left lane change, right lane change, and keeping straight.

[0018] Step 2.2. Lane Change Feasibility Determination: If the intention to maintain straight-ahead is recognized, the co-driving system does not output lateral control. For left and right lane changes, the feasibility of the lane change is first determined. In the geodetic coordinate system, the vehicle is assumed to be a rectangle. If the vehicle's lane-changing process with maximum lateral acceleration has no intersection with other rectangles, then the lane change feasibility condition is met. When lane change feasibility is met, the co-driving system generates a lane-changing trajectory for lane-changing assistance. The following are the cases where lane change feasibility is not met:

[0019] Scenario 1: The driver fails to effectively assess the situation of vehicles behind in the target lane and changes lanes too early. In this case, the longitudinal distance between the driver and the vehicles behind in the target lane is too small, posing a risk of collision.

[0020] Scenario 2: The vehicle in front of you in the current lane brakes suddenly, and your vehicle fails to take timely and effective lane-changing avoidance measures, resulting in a risk of collision due to insufficient longitudinal distance between you and the vehicle in front of you in the original lane or the vehicle in front of you in the target lane.

[0021] As a preferred embodiment of the present invention, step 2 further includes the following step:

[0022] Step 2.3. For case 1 where lane changing is not feasible in step 2.2, the co-driving system generates the desired straight-ahead trajectory and takes over vehicle control in a timely manner to ensure safety.

[0023] Step 2.4. In case 2 where lane change feasibility is not met in step 2.2, the co-driving system promptly takes over vehicle control to ensure safety, controls the vehicle to brake and decelerate to make room for subsequent lane change actions, so that when the driver's intention to change lanes is detected again, the lane change feasibility judgment can be met, and a safe expected lane change trajectory can be generated.

[0024] As a preferred embodiment of the present invention, step 2 further includes the following step:

[0025] Step 2.5. Trajectory Cluster Construction: When lane-changing feasibility is satisfied, the desired trajectory is designed with longitudinal constant speed and lateral fifth-order polynomial. Multiple repeated tests are conducted under four lane-changing scenarios: the maximum speed of the vehicle ahead in the current lane is 15 m / s, 20 m / s, 25 m / s, and 30 m / s, and the maximum braking deceleration is -4 m / s². The lane-changing time T for each scenario is recorded. lc Mean characteristics and standard deviation characteristics by As the lower boundary, with With 0.2s as the upper bound and 0.2s as the step size, the generated series of lane-changing times ultimately construct a two-dimensional trajectory cluster for each lane-changing scenario.

[0026] As a preferred embodiment of the present invention, step 2 further includes the following step:

[0027] Step 2.6. Solving for the optimal lane-changing trajectory; A multi-objective scoring method is used to select the trajectory with the highest score as the optimal lane-changing trajectory for the current scenario. The multi-objective scoring considers the following four sub-objectives: stability objective, efficiency objective, comfort objective, and anthropomorphism objective. The results are as follows:

[0028] score = k1score s +k2score e +k3score c +k4score h ;

[0029] In the formula, score is the final score of the compared trajectory. s score e score c score h

[0030] The scores are for stability, efficiency, comfort, and anthropomorphism objectives, respectively. k1, k2, k3, and k4 are the weights corresponding to the scores of each sub-objective. The calculation methods for each indicator score are as follows:

[0031]

[0032] In the formula, a y_max Let g be the maximum lateral acceleration for each planned trajectory, g be the acceleration due to gravity, and T be the acceleration due to gravity. lc The lane change time required for the evaluated trajectory, i.e., the lane change time statistically analyzed for each scenario. The longest lane-changing time under the planned trajectory is... j y_max j is the maximum lateral jerk under the planned trajectory. y_ref This is a reference value for lateral acceleration. The lane change time T is calculated for each scenario. lc The mean characteristics, The lane change time T is calculated for each scenario. lc The standard deviation characteristic.

[0033] As a preferred embodiment of the present invention, step 3 further includes the following step:

[0034] Step 3.1. When the driver's intention recognition result is to maintain straight driving, the driver's weight is set to 1, and the weight of the co-driving system is set to 0;

[0035] Step 3.2. When the driver's intent recognition result is a left lane change or a right lane change, and the lane change feasibility conditions are not met, the driver's weight is set to 0, and the weight of the co-driving system is set to 1.

[0036] Step 3.3. When the driver's intention recognition result is a left lane change or a right lane change, and the lane change feasibility condition is met, firstly, using a rolling time window of 0.5s, calculate the driver's steering wheel angle δ in each window. f The standard deviation of the steering wheel angle is used as the basis for constructing a log-normal distribution based on the statistical results of the logarithmic values ​​of the standard deviation of the steering wheel angle. The probability density function f(x; μ, δ) is:

[0037]

[0038] In the formula, x is the standard deviation of the steering wheel angle, μ and δ are the mean and standard deviation of the logarithm of the standard deviation of the steering wheel angle, and its cumulative distribution function is expressed as:

[0039]

[0040] In human-machine co-driving, the real-time weight a of the driver and the co-driving system within the current time window i. h (i) and the real-time weights of the co-driving system a s (i) is defined as follows:

[0041]

[0042] In the formula, a h (i) represents the driver's weight within time window i-1. The weight is assigned when the driver's initial weight at the start of a lane change is 1, and the change in driver's weight between adjacent time windows is no greater than 0.2. Otherwise, the driver's weight value is a. s (i) is 1 minus the cumulative distribution function value of the standard deviation of the steering wheel angle under the corresponding window.

[0043] As a preferred embodiment of the present invention, step 4 further includes the following step:

[0044] Step 4.1. Based on the optimal lane-changing trajectory obtained by the solution, apply the adaptive model predictive control method to track the desired trajectory. Define the model state space d(ξ) / d(t)=f(ξ,u), where the state variable ξ is defined as follows: Among them, v y v x For lateral velocity and longitudinal velocity, and Let yaw angle and yaw rate be the vehicle's yaw angle and yaw rate, and Y and X be the vehicle's lateral and longitudinal positions in the geodetic coordinate system, respectively. The control input u and output η are defined as follows: u = [σ f ],

[0045] As a preferred embodiment of the present invention, step 4 further includes the following step:

[0046] Step 4.2. Based on the adaptive model predictive control method established in Step 4.1, the co-driving system constructs a quadratic programming optimization problem with control increment and output deviation as objectives. The objective function of this problem comprehensively considers the vehicle's trajectory tracking accuracy and control smoothness, and is defined as follows:

[0047] The control sequence is solved by minimizing the sum of squared deviations between the expected trajectory and the actual output in the predicted time domain and the weighted sum of squared control input increments, while satisfying vehicle dynamics constraints, control quantity constraints and control increment constraints.

[0048] The quadratic programming problem is expressed as:

[0049]

[0050] Among them, Y ref,i To predict the desired output in the time domain, Y i For the model's predicted output, ΔU j Let Q be the control increment at step j, and R be the weighted matrices of the output error and the control increment, respectively. Let N be the control increment at step j. p To predict the time domain length, N c To control the length of the time domain;

[0051] The co-driving system uses the effective set method to solve the quadratic programming problem in real time, obtaining a series of control input increments and relaxation factors in the control time domain. The co-driving system uses the first control quantity in the control sequence as the actual control input at the current moment to act on the vehicle, realizing dynamic tracking of the desired lane-changing trajectory; then it enters the next moment for rolling optimization until the entire trajectory tracking and control process is completed.

[0052] As a preferred embodiment of the present invention, step 4 further includes the following step:

[0053] Step 4.3. The adaptability of the adaptive model predictive control method is used in the control time domain N c and prediction time domain N p The relationship between longitudinal vehicle speed and speed is set as follows:

[0054]

[0055] In the formula, k is the proportionality coefficient, and v xmin v xmax These are the longitudinal low-speed threshold and the longitudinal high-speed threshold, respectively, N pmax To predict the upper limit of the time domain.

[0056] Advantages and benefits of the present invention:

[0057] (1) The method provided by the present invention integrates the LSTM model and the CNN model for driver intent recognition, and integrates temporal features and spatial features to ensure a high recognition accuracy.

[0058] (2) The method provided by this invention is based on the results of driver lane-changing intention recognition and lane-changing feasibility condition determination, and finally determines whether and how the co-driving system should provide assistance. In lane-changing assistance, multi-objective dynamic weight allocation is set to further reduce human-machine conflict and improve the driving experience.

[0059] (3) The method provided by the present invention dynamically adjusts the weight between the human and the machine by constructing a functional relationship between the weight allocation value between the human and the machine and the standard deviation of the steering wheel angle of the driver in the window, while ensuring that the weight fluctuation is not too large, thus ensuring both safety and driving comfort.

[0060] (4) The method provided by the present invention sets the prediction time domain and control time domain in model predictive control as functions of longitudinal velocity, which effectively ensures the model solution speed and trajectory tracking accuracy.

[0061] (5) The method provided by this invention is mainly divided into four parts: intent recognition, trajectory planning, permission allocation and motion control. Each part is relatively independent but organically connected. Under the premise of ensuring safety, it can provide personalized lane change assistance for different drivers and effectively reduce human-machine conflict. Attached Figure Description

[0062] Other objects and results of the invention will become more apparent and readily understood with reference to the following description taken in conjunction with the accompanying drawings. In the drawings:

[0063] Figure 1 This is a logic block diagram provided in this embodiment;

[0064] Figure 2This is a schematic diagram illustrating the critical collision moment in the lane-changing feasibility determination provided in this embodiment, where a lane change to the left is likely to result in a minor collision with the vehicle in front of the original lane.

[0065] Figure 3 This is a schematic diagram illustrating the critical collision moment when changing lanes to the left and the vehicle in front of the target lane is likely to collide during the lane change feasibility determination provided in this embodiment.

[0066] Figure 4 This is a schematic diagram illustrating the critical collision moment when a lane change to the left is likely to result in a minor collision with a vehicle behind the target lane, as provided in this embodiment for determining the feasibility of lane changing.

[0067] Figure 5 A flowchart of the trajectory planning decision scheme provided in this embodiment. Detailed Implementation

[0068] A method for permission allocation and planning control in a co-driving system that considers driving intentions and states includes the following steps:

[0069] Step 1. Driver intention recognition model construction: Through multimodal data collection (such as vehicle status and environmental images), spatial and temporal features are extracted by combining CNN and LSTM models, and the driving intention (left lane change, right lane change or straight driving) is recognized in real time after fusion.

[0070] Step 1.1. Real-time multimodal data acquisition: During vehicle operation, data is collected in real time using various sensors: (1) Time series data: Real-time status is obtained from the vehicle's CAN (Controller Area Network) bus, including position, speed, acceleration, steering wheel angle, and light signals; (2) Spatial feature data: Real-time images, lane lines, and relative positions of surrounding vehicles are obtained through environmental perception devices. The time series data is sliced ​​according to a fixed time window (data from the past second, sampled 10 times per second) to form a sliding time window; at the same time, the spatial feature data is dynamically updated frame by frame to match the time series data. The environmental perception devices include cameras, radar, or lidar.

[0071] Step 1.2. Construct a CNN (Convolutional Neural Network) model to process spatial features; preprocess the input image data by scaling and normalizing to adapt to the input format of the CNN model, use convolutional layers to extract local spatial features, including lane line shapes and the relative positions of vehicles in front or to the side, use pooling layers to reduce dimensionality, retain the preset key features, and reduce the amount of computation to meet real-time requirements, and finally obtain the spatial feature vector;

[0072] Step 1.3. Construct an LSTM (Long Short-Term Memory) model to process time series features. According to the sliding time window, input the time series data of the most recent time (the past 1 second) into the LSTM model. Use multiple LSTM modules of the LSTM model to process the time series in sequence, capture the dynamic change trend of driving behavior, and output the time series feature vector to represent the continuous dynamic features of the vehicle.

[0073] Step 1.4. Real-time feature fusion: At the end of each time window, the features extracted by the LSTM model and the CNN model are fused. The time series feature vector output by the CNN model and the spatial feature vector output by the CNN model are combined by feature concatenation. The fused feature vector is then input into a fully connected layer for further processing and compression to form high-level semantic features.

[0074] Step 1.5. Classify and output the recognition results. Based on the fused high-level semantic features, predict driving intentions in real time. Use a fully connected layer to map the high-level semantic features to the classification space. Calculate the probability of each driving intention category through the Softmax activation function. The output driving intention categories include: left lane change, right lane change, and keep straight.

[0075] Step 2. Trajectory planning decision design: Based on the identified driving intention, determine the feasibility of lane changing, generate a safe trajectory cluster, and use a multi-objective scoring method to select the optimal lane changing trajectory to achieve efficient, safe, and human-like trajectory planning;

[0076] Step 2.1. Input the intent recognition result: Input the driving intent category output by the driver intent recognition model into the trajectory planning decision module. The driving intent categories include left lane change, right lane change, and keeping straight.

[0077] Step 2.2. Lane Change Feasibility Determination: If the intention to continue straight is recognized, the co-driving system does not output lateral control. If the intention to change lanes to the left or right is recognized, a lane change feasibility determination is first performed. In the geodetic coordinate system (XOY), the vehicle is assumed to be a rectangle. If the vehicle changes lanes with maximum lateral acceleration and there are no intersections with other rectangles (vehicles), the lane change feasibility condition is met. When lane change feasibility is met, the co-driving system generates a lane change trajectory for lane change assistance. The following are the cases where lane change feasibility is not met:

[0078] Scenario 1: The driver fails to effectively assess the situation of vehicles behind in the target lane and changes lanes too early. In this case, the longitudinal distance between the driver and the vehicles behind in the target lane is too small, posing a risk of collision.

[0079] Scenario 2: The vehicle in front of you in the current lane brakes suddenly, and your vehicle fails to take timely and effective lane-changing avoidance measures, resulting in a risk of collision due to insufficient longitudinal distance between you and the vehicle in front of you in the original lane or the vehicle in front of you in the target lane.

[0080] The calculation of lane-changing feasibility conditions is as follows: Given the coordinates of the center of gravity as (X, Y), the first step is to solve for the yaw angle at the vehicle's center of gravity.

[0081]

[0082] In the formula, v y v is the lateral velocity at the vehicle's center of mass. x The longitudinal velocity at the vehicle's center of gravity.

[0083] There are three different dangerous collision scenarios that can occur when changing lanes to the left. The dangerous scenarios for changing lanes to the right are the same as those for changing lanes to the left, so we will only consider the different dangerous scenarios for changing lanes to the left here:

[0084] The first type of lane change to the left is prone to collisions with vehicles in the original lane. The upper right vertex of the vehicle's rectangle is most likely to collide with the rear left vertex of the vehicle in front. The formula for calculating the coordinates (X1, Y1) of the vehicle's right front vertex during the lane change is:

[0085]

[0086] In the formula, l a l is the distance from the center of mass to the front of the car. c Let (X, Y) be the distance from the center of mass to the sides of the car, and (X, Y) be the coordinates of the center of mass in the geodetic coordinate system. Let the coordinates of the left rear vertex of the rectangle of the car in front of the original lane be (X2, Y2). It can be seen that when the car travels to (X1, Y1) with the maximum lateral acceleration under the current road conditions, if the lateral position Y1 of the right front vertex of the car is greater than the lateral position Y2 of the left rear vertex of the car in front, then the car will not collide with the car in front during the entire lane change process. Therefore, the lane change trajectory satisfies the lane change feasibility condition.

[0087] The second type of lane change to the left is prone to collisions with vehicles in the target lane. The front left vertex of the vehicle's rectangle is most likely to collide with the rear right vertex of the vehicle in front. The formula for calculating the coordinates (X3, Y3) of the vehicle's top left vertex during the lane change is as follows:

[0088]

[0089] In the formula, let the coordinates of the right rear vertex of the rectangle of the vehicle in front of the target lane be (X4, Y4). It can be seen that when the vehicle travels to X3 = Y4 with the maximum lateral acceleration under the current road conditions, if the lateral position Y3 of the left front vertex of the vehicle is less than the lateral position Y4 of the right rear vertex of the vehicle in front, then the vehicle will not collide with the vehicle in front during the entire lane change process, and the lane change trajectory satisfies the lane change feasibility condition.

[0090] The third scenario involves a lane change to the left, where a collision with a vehicle behind in the target lane is likely. The left rear vertex of the vehicle's rectangle is most likely to collide with the right front vertex of the vehicle behind. The formula for calculating the coordinates (X5, Y5) of the left rear vertex of the vehicle during the lane change is as follows:

[0091]

[0092] In the formula, l b Let X6 be the distance from the center of mass to the rear of the car. Let the coordinates of the right front vertex of the rectangle of the target lane be (X6, Y6). It can be seen that when the car travels to Y5 = Y6 with the maximum lateral acceleration under the current road conditions, if the lateral position X5 of the left rear vertex of the car is greater than the lateral position X6 of the right front vertex of the car, then the car will not collide with the car in the entire lane change process. Therefore, the lane change trajectory satisfies the lane change feasibility condition.

[0093] Step 2.3. For case 1 where lane changing is not feasible in step 2.2, the co-driving system generates the desired straight-ahead trajectory and takes over vehicle control in a timely manner to ensure safety.

[0094] Step 2.4. In case 2 where lane change feasibility is not met in step 2.2, the co-driving system promptly takes over vehicle control to ensure safety, controls the vehicle to brake and decelerate to make room for subsequent lane change actions, so that when the driver's intention to change lanes is detected again, the lane change feasibility judgment can be met, and a safe expected lane change trajectory can be generated.

[0095] The selection of the maximum lateral acceleration is related to the road surface conditions. Under good road surface conditions, the maximum lateral acceleration is set to 0.4g, while under general road surface conditions, the maximum lateral acceleration is set to μg, where μ is the current road surface adhesion coefficient.

[0096] The design of the early lane-changing trajectory for scenario 2 is as follows: First, calculate the lane-changing time T of the set trajectory. x Given a uniform longitudinal velocity and a fifth-degree polynomial lateral velocity, the simplest trajectory equation is:

[0097]

[0098] In the formula, v desLet w be the desired longitudinal velocity, w be the lateral displacement required for lane changing (generally set to the standard road width), X be the vehicle's longitudinal position (i.e., the longitudinal position in the geodetic coordinate system), and D be the longitudinal displacement required for lane changing. Then, taking the derivative of Y(t) twice with respect to time t yields the lateral acceleration a. y The equation for (t) can be obtained by applying the chain rule:

[0099]

[0100] Let a y (t)=a y (T x (2 / 2) = 0.4g, calculate the corresponding lane-changing time T. x And the longitudinal displacement d required for lane changing col The obtained d col Substitute into Y (t) In the middle, let X1 = X2 and Y1 = Y2, and finally obtain the corresponding critical collision time t. col If the driver changes lanes too late, the co-driving system will generate a timeframe corresponding to the critical collision time t. col If the lane change trajectory is adjusted 0.5 seconds in advance, the conditions for lane change feasibility are met, ensuring safety.

[0101] Step 2.5. Trajectory Cluster Construction: When lane-changing feasibility is satisfied, the desired trajectory is designed with longitudinal constant speed and lateral fifth-order polynomial. Multiple repeated tests are conducted under four lane-changing scenarios: the maximum speed of the vehicle ahead in the current lane is 15 m / s, 20 m / s, 25 m / s, and 30 m / s, and the maximum braking deceleration is -4 m / s². The lane-changing time T for each scenario is recorded. lc Mean characteristics and standard deviation characteristics by As the lower boundary, with With 0.2s as the upper bound and 0.2s as the step size, the generated series of lane-changing times ultimately construct a two-dimensional trajectory cluster for each lane-changing scenario.

[0102] Step 2.6. Solving for the optimal lane-changing trajectory; A multi-objective scoring method is used to select the trajectory with the highest score as the optimal lane-changing trajectory for the current scenario. The multi-objective scoring considers the following four sub-objectives: stability objective, efficiency objective, comfort objective, and anthropomorphism objective. The results are as follows:

[0103] score = k1score s +k2score e +k3score c +k4score h ;

[0104] In the formula, score is the final score of the compared trajectory.s score e score c score h

[0105] The scores are for stability, efficiency, comfort, and anthropomorphism objectives, respectively. k1, k2, k3, and k4 are the weights corresponding to the scores of each sub-objective. The calculation methods for each indicator score are as follows:

[0106]

[0107] In the formula, a y_max Let g be the maximum lateral acceleration for each planned trajectory, g be the acceleration due to gravity, and T be the acceleration due to gravity. lc The lane change time required for the evaluated trajectory, i.e., the lane change time statistically analyzed for each scenario. The longest lane-changing time under the planned trajectory is... j y_max j is the maximum lateral jerk under the planned trajectory. y_ref This is a reference value for lateral acceleration. The lane change time T is calculated for each scenario. lc The mean characteristics, The lane change time T is calculated for each scenario. lc The standard deviation characteristic.

[0108] The determination of the weighting coefficients k1, k2, k3, and k4 for the stability, efficiency, comfort, and anthropomorphism objectives mainly considers the influence of vehicle speed, traffic density, and road conditions.

[0109] In high-speed, low-density scenarios, efficiency takes precedence, ensuring rapid lane changes. Stability is also important, but comfort and human-like characteristics have lower weights, focusing primarily on quick and safe lane changes. The weighting is as follows: Stability: 30%, Efficiency: 50%, Comfort: 10%, Human-like Characteristics: 10%. In low-speed, high-density, and complex road conditions, stability and comfort are paramount, ensuring safe and smooth operation. Efficiency has a lower weight, while human-like characteristics are moderately increased to make the system's operation appear more natural. The weighting is as follows: Stability: 35%, Efficiency: 20%, Comfort: 30%, Human-like Characteristics: 15%. In medium-speed, moderate-density environments, the weights of the four objectives are more balanced. The system needs to balance efficiency, comfort, and stability while ensuring operation meets the driver's expectations and exhibits certain human-like characteristics. The weighting is as follows: Stability: 30%, Efficiency: 20%, Comfort: 25%, Human-like Characteristics: 25%.

[0110] This dynamic weighting method allows the system to flexibly adjust in different driving scenarios, balancing safety, comfort, and efficiency, while simulating human driver behavior to provide a more natural driving experience.

[0111] Step 3. Real-time solution of optimal human-machine weights: Based on the driver's intentions and steering wheel angle fluctuations, dynamically adjust the human-machine weight allocation to achieve smooth collaboration and safe switching between the driver and the co-driving system.

[0112] Step 3.1. When the driver's intention recognition result is to maintain straight driving, the driver's weight is set to 1, and the weight of the co-driving system is set to 0;

[0113] Step 3.2. When the driver's intention recognition result is a left lane change or a right lane change, and the lane change feasibility conditions are not met, i.e., in cases 1 and 2 in step 2.2, the driver's weight is set to 0, and the weight of the co-driving system is set to 1.

[0114] Step 3.3. When the driver's intention recognition result is a left lane change or a right lane change, and the lane change feasibility condition is met, firstly, using a rolling time window of 0.5s, calculate the driver's steering wheel angle δ in each window. f The standard deviation of the steering wheel angle is used as the basis for constructing a log-normal distribution based on the statistical results of the logarithmic values ​​of the standard deviation of the steering wheel angle. The probability density function f(x; μ, δ) is:

[0115]

[0116] In the formula, x is the standard deviation of the steering wheel angle, μ and δ are the mean and standard deviation of the logarithm of the standard deviation of the steering wheel angle, and its cumulative distribution function is expressed as:

[0117]

[0118] As shown in the above formula, the larger the standard deviation of the steering wheel angle within the window, the larger the value of the corresponding cumulative distribution function, and the smaller the corresponding driver's weight should be. At the same time, considering that the change in permissions between adjacent time windows should not be too large, the real-time weight a of the driver and the co-driving system in the current time window i under human-machine co-driving should be... h (i) and the real-time weights of the co-driving system a s (i) is defined as follows:

[0119]

[0120] In the formula, a h (i) represents the driver's weight within time window i-1. The weight is assigned when the driver's initial weight at the start of a lane change is 1, and the change in driver's weight between adjacent time windows is no greater than 0.2. Otherwise, the driver's weight value is a. s(i) is 1 minus the cumulative distribution function value of the standard deviation of the steering wheel angle under the corresponding window.

[0121] Step 4. Solve for motion control variables. Based on the optimal lane-changing trajectory, adopt the adaptive model predictive control (MPC) algorithm, and combine the vehicle speed dynamic adjustment prediction and control time domain to achieve high-precision tracking control of the vehicle on the desired trajectory.

[0122] Step 4.1. Based on the optimal lane-changing trajectory obtained by the solution, apply the adaptive model predictive control method to track the desired trajectory. Define the model state space d(ξ) / d(t)=f(ξ,u), where the state variable ξ is defined as follows: Among them, v y v x For lateral velocity and longitudinal velocity, and Let yaw angle and yaw rate be the vehicle's yaw angle and yaw rate, and Y and X be the vehicle's lateral and longitudinal positions in the geodetic coordinate system (XOY). The control input u and output η are defined as follows: u = [σ f ],

[0123] Step 4.2. Based on the adaptive model predictive control method (adaptive model predictive control framework) established in Step 4.1, the co-driving system constructs a quadratic programming (QP) optimization problem with control increment and output deviation as objectives. The objective function of this problem comprehensively considers the vehicle's trajectory tracking accuracy and control smoothness, and is defined as follows:

[0124] The optimal control sequence is solved by minimizing the sum of squared deviations between the expected trajectory and the actual output in the predicted time domain and the weighted sum of squared control input increments, while satisfying vehicle dynamics constraints, control quantity constraints, and control increment constraints.

[0125] Specifically, the quadratic programming problem is expressed as:

[0126]

[0127] Among them, Y ref,i To predict the desired output in the time domain, Y i For the model's predicted output, ΔU j Let Q be the control increment at step j, and R be the weighted matrices of the output error and the control increment, respectively. Let N be the control increment at step j. p To predict the time domain length, N c To control the length of the time domain;

[0128] The co-driving system uses the Active Set Method to solve the quadratic programming problem in real time, obtaining a series of optimal control input increments and relaxation factors in the control time domain. The co-driving system uses the first control variable in the control sequence as the actual control input at the current moment to apply to the vehicle, thereby achieving dynamic tracking of the desired lane-changing trajectory. Then, it enters the next moment for rolling optimization until the entire trajectory tracking and control process is completed.

[0129] Step 4.3. The adaptability of the adaptive model predictive control method is used in the control time domain N c and prediction time domain N p The relationship between longitudinal vehicle speed and speed is set as follows:

[0130]

[0131] In the formula, k is the proportionality coefficient, and v xmin v xmax These are the longitudinal low-speed threshold and the longitudinal high-speed threshold, respectively, N pmax To predict the upper limit of the time domain.

Claims

1. A method for permission allocation and planning control in a co-driving system that considers driving intentions and states, characterized in that, Includes the following steps: Step 1. Driver intent recognition model construction: Through multimodal data collection, spatial and temporal features are extracted by combining CNN and LSTM models, and the driving intent is recognized in real time after fusion. Step 2. Trajectory planning decision design: Based on the identified driving intention, determine the feasibility of lane changing, generate a safe trajectory cluster, and use a multi-objective scoring method to select the optimal lane changing trajectory to achieve trajectory planning; Step 3. Real-time solution of optimal human-machine weights: Based on the driver's intentions and steering wheel angle fluctuations, dynamically adjust the human-machine weight allocation to achieve smooth collaboration and safe switching between the driver and the co-driving system. Step 3.

1. When the driver's intention recognition result is to maintain straight driving, the driver's weight is set to 1, and the weight of the co-driving system is set to 0; Step 3.

2. When the driver's intent recognition result is a left lane change or a right lane change, and the lane change feasibility conditions are not met, the driver's weight is set to 0, and the weight of the co-driving system is set to 1. Step 3.

3. When the driver's intention recognition result is a left lane change or a right lane change, and the lane change feasibility condition is met, firstly, using a rolling time window of 0.5s, calculate the driver's steering wheel angle δ in each window. f The standard deviation of the steering wheel angle is used as the basis for constructing a log-normal distribution based on the statistical results of the logarithmic values ​​of the standard deviation of the steering wheel angle. The probability density function f(x; μ, δ) is: In the formula, x is the standard deviation of the steering wheel angle, μ and δ are the mean and standard deviation of the logarithm of the standard deviation of the steering wheel angle, and its cumulative distribution function is expressed as: In human-machine co-driving, the real-time weight a of the driver and the co-driving system within the current time window i. h (i) and the real-time weights of the co-driving system a s (i) is defined as follows: In the formula, a h (i) represents the driver's weight within time window i-1. The weight is assigned when the driver's initial weight at the lane change is 1, and the change in weight between adjacent time windows is no greater than 0.

2. Otherwise, the driver's weight value is a. s (i) is 1 minus the cumulative distribution function value of the standard deviation of the steering wheel angle under the corresponding window; Step 4. Solve for motion control variables. Based on the optimal lane-changing trajectory, adopt an adaptive model predictive control method, and combine vehicle speed dynamic adjustment prediction and control time domain to achieve high-precision tracking control of the vehicle on the desired trajectory.

2. The method for permission allocation and planning control of a co-driving system considering driving intention and state according to claim 1, characterized in that, Step 1 also includes the following steps: Step 1.

1. Real-time multimodal data acquisition: During vehicle operation, data is collected in real time using sensors: (1) Time series data: real-time status is obtained from the vehicle's CAN bus, including position, speed, acceleration, steering wheel angle, and light signals; (2) Spatial feature data: real-time images, lane lines, and relative positions of surrounding vehicles are obtained through environmental perception devices. The time series data is sliced ​​according to a fixed time window to form a sliding time window; at the same time, the spatial feature data is dynamically updated frame by frame to match the time series data. The environmental perception devices include cameras, radar, or lidar. Step 1.

2. Construct a CNN model to process spatial features; preprocess the input image data by scaling and normalizing to adapt to the input format of the CNN model, use convolutional layers to extract local spatial features, including lane line shape, relative position of vehicles in front or on the side, use pooling layers to reduce dimensionality, retain the preset key features, and reduce the amount of computation to meet real-time requirements, and finally obtain spatial feature vectors. Step 1.

3. Construct an LSTM model to process time series features. According to the sliding time window, input the time series data of the most recent time into the LSTM model. Use multiple LSTM modules of the LSTM model to process the time series in sequence, capture the dynamic change trend of driving behavior, and output the time series feature vector to represent the continuous dynamic features of the vehicle. Step 1.

4. Real-time feature fusion: At the end of each time window, the features extracted by the LSTM model and the CNN model are fused. The time series feature vector output by the CNN model and the spatial feature vector output by the CNN model are combined by feature concatenation. The fused feature vector is input into the fully connected layer for further processing and compression to form high-level semantic features. Step 1.

5. Classify and output the recognition results. Based on the fused high-level semantic features, predict driving intentions in real time. Use a fully connected layer to map the high-level semantic features to the classification space. Calculate the probability of each driving intention category through the Softmax activation function. The output driving intention categories include: left lane change, right lane change, and keep straight.

3. The method for permission allocation and planning control of a co-driving system considering driving intention and state according to claim 1, characterized in that, Step 2 also includes the following steps: Step 2.

1. Input the intent recognition result: Input the driving intent category output by the driver intent recognition model into the trajectory planning decision module. The driving intent categories include left lane change, right lane change, and keeping straight. Step 2.

2. Lane Change Feasibility Determination: If the intention to maintain straight-ahead is recognized, the co-driving system does not output lateral control. For left and right lane changes, the feasibility of the lane change is first determined. In the geodetic coordinate system, the vehicle is assumed to be a rectangle. If the vehicle's lane-changing process with maximum lateral acceleration has no intersection with other rectangles, then the lane change feasibility condition is met. When lane change feasibility is met, the co-driving system generates a lane-changing trajectory for lane-changing assistance. The following are the cases where lane change feasibility is not met: Scenario 1: The driver fails to effectively assess the situation of vehicles behind in the target lane and changes lanes too early. In this case, the longitudinal distance between the driver and the vehicles behind in the target lane is too small, posing a risk of collision. Scenario 2: The vehicle in front of you in the current lane brakes suddenly, and your vehicle fails to take timely and effective lane-changing avoidance measures, resulting in a risk of collision due to insufficient longitudinal distance between you and the vehicle in front of you in the original lane or the vehicle in front of you in the target lane.

4. The method for permission allocation and planning control of a co-driving system considering driving intention and state according to claim 3, characterized in that, Step 2 also includes the following steps: Step 2.

3. For case 1 where lane changing is not feasible in step 2.2, the co-driving system generates the desired straight-ahead trajectory and takes over vehicle control in a timely manner to ensure safety. Step 2.

4. In case 2 where lane change feasibility is not met in step 2.2, the co-driving system promptly takes over vehicle control to ensure safety, controls the vehicle to brake and decelerate to make room for subsequent lane change actions, so that when the driver's intention to change lanes is detected again, the lane change feasibility judgment can be met, and a safe expected lane change trajectory can be generated.

5. The method for permission allocation and planning control of a co-driving system considering driving intention and state according to claim 4, characterized in that, Step 2 also includes the following steps: Step 2.

5. Trajectory Cluster Construction: When lane-changing feasibility is satisfied, the desired trajectory is designed with longitudinal constant speed and lateral fifth-order polynomial. Multiple repeated tests are conducted under four lane-changing scenarios: the maximum speed of the vehicle ahead in the current lane is 15 m / s, 20 m / s, 25 m / s, and 30 m / s, and the maximum braking deceleration is -4 m / s². The lane-changing time T for each scenario is recorded. lc Mean characteristics and standard deviation characteristics by As the lower boundary, with With 0.2s as the upper bound and 0.2s as the step size, the generated series of lane-changing times ultimately construct a two-dimensional trajectory cluster for each lane-changing scenario.

6. The method for permission allocation and planning control of a co-driving system considering driving intention and state according to claim 5, characterized in that, Step 2 also includes the following steps: Step 2.

6. Solving for the optimal lane-changing trajectory; A multi-objective scoring method is used to select the trajectory with the highest score as the optimal lane-changing trajectory for the current scenario. The multi-objective scoring considers the following four sub-objectives: stability objective, efficiency objective, comfort objective, and anthropomorphism objective. The results are as follows: score=k1score s +k2score e +k3score c +k4score h ; In the formula, score is the final score of the compared trajectory. s score e score c score h The scores are for stability, efficiency, comfort, and anthropomorphism objectives, respectively. k1, k2, k3, and k4 are the weights corresponding to the scores of each sub-objective. The calculation methods for each indicator score are as follows: In the formula, a y_max Let g be the maximum lateral acceleration for each planned trajectory, g be the acceleration due to gravity, and T be the acceleration due to gravity. lc The lane change time required for the evaluated trajectory, i.e., the lane change time statistically analyzed for each scenario. The longest lane-changing time under the planned trajectory is... j y_max j is the maximum lateral jerk under the planned trajectory. y_ref This is a reference value for lateral acceleration. The lane change time T is calculated for each scenario. lc The mean characteristics, The lane change time T is calculated for each scenario. lc The standard deviation characteristic.

7. The method for permission allocation and planning control of a co-driving system considering driving intention and state according to claim 1, characterized in that, Step 4 also includes the following steps: Step 4.

1. Based on the optimal lane-changing trajectory obtained by the solution, apply the adaptive model predictive control method to track the desired trajectory. Define the model state space d(ξ) / d(t)=f(ξ,u), where the state variable ξ is defined as follows: Among them, v y v x For lateral velocity and longitudinal velocity, and Let yaw angle and yaw rate be the vehicle's yaw angle and yaw rate, and Y and X be the vehicle's lateral and longitudinal positions in the geodetic coordinate system, respectively. The control input u and output η are defined as follows: u = [σ f ], 8. The method for permission allocation and planning control of a co-driving system considering driving intention and state according to claim 7, characterized in that, Step 4 also includes the following steps: Step 4.

2. Based on the adaptive model predictive control method established in Step 4.1, the co-driving system constructs a quadratic programming optimization problem with control increment and output deviation as objectives. The objective function of this problem comprehensively considers the vehicle's trajectory tracking accuracy and control smoothness, and is defined as follows: The control sequence is solved by minimizing the sum of squared deviations between the expected trajectory and the actual output in the predicted time domain and the weighted sum of squared control input increments, while satisfying vehicle dynamics constraints, control quantity constraints and control increment constraints. The quadratic programming problem is expressed as: Among them, Y ref,i To predict the desired output in the time domain, Y i For the model's predicted output, ΔU j Let Q be the control increment at step j, and R be the weighted matrices of the output error and the control increment, respectively. Let N be the control increment at step j. p To predict the time domain length, N c To control the length of the time domain; The co-driving system uses the effective set method to solve the quadratic programming problem in real time, obtaining a series of control input increments and relaxation factors in the control time domain. The co-driving system uses the first control quantity in the control sequence as the actual control input at the current moment to act on the vehicle, realizing dynamic tracking of the desired lane-changing trajectory; then it enters the next moment for rolling optimization until the entire trajectory tracking and control process is completed.

9. A method for permission allocation and planning control of a co-driving system considering driving intention and state, as described in claim 7, is characterized in that... Step 4 also includes the following steps: Step 4.

3. The adaptability of the adaptive model predictive control method is used in the control time domain N c and prediction time domain N p The relationship between longitudinal vehicle speed and speed is set as follows: N c =0.3N p , In the formula, k is the proportionality coefficient, and v xmin v xmax These are the longitudinal low-speed threshold and the longitudinal high-speed threshold, respectively, N pmax To predict the upper limit of the time domain.

Images