Lane changing decision-making method and system for automatic driving vehicle and automatic driving vehicle

Abstract

The invention belongs to the technical field of automobiles, and provides an automatic driving vehicle lane changing decision-making method and system and an automatic driving vehicle. According to the technical scheme, obtained road environment information is preprocessed; training the constructed spatial-temporal feature extraction network based on the preprocessed road environment information to obtain a trained spatial-temporal feature extraction network; combining the driving environment data sensed in real time and the trained feature extraction network to predict and obtain an environment sensing result and a lane changing decision prediction result; fusing an environment perception result and a lane changing decision prediction result to obtain state representation, and adjusting an optimal lane changing strategy in combination with the state representation and an action value function; and performing path planning based on the adjusted optimal lane changing strategy, and generating a steering wheel angle signal and a vehicle speed control signal for controlling the vehicle according to a path planning result. And a more accurate lane changing decision can be realized.

Description

Technical Field

[0001] The present invention belongs to the field of vehicle assisted driving, and in particular relates to a lane changing decision method and system for an autonomous driving vehicle and an autonomous driving vehicle. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Autonomous driving technology aims to achieve autonomous navigation and control of vehicles through computer vision, machine learning and other technologies. Lane changing decision is a key link in autonomous driving, which requires the vehicle to accurately perceive the surrounding environment, predict the behavior of other vehicles and pedestrians, and make appropriate lane changing decisions accordingly.

[0004] However, the road traffic environment is extremely complex and changeable, including different road conditions, traffic flow, weather conditions, and the behavior of pedestrians and other vehicles. Traditional lane change decision methods, such as rule-based methods or simple machine learning methods, often fail to fully consider these complex factors and may perform poorly in the face of complex scenarios. Summary of the invention

[0005] In view of this, the present invention aims to propose a lane changing decision method for an autonomous driving vehicle, which learns complex driving strategies and feature representations from a large amount of driving data, and in the face of different complex situations, such as situations involving obstacles or traffic jams, achieves accurate vehicle lane changing by adjusting the steering wheel angle signal and the vehicle speed control signal in real time, thereby reducing various risks encountered during the lane changing process and ensuring safe and efficient driving.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A lane-changing decision method for an autonomous driving vehicle comprises the following steps:

[0008] Preprocessing the acquired road environment information;

[0009] The constructed spatiotemporal feature extraction network is trained based on the preprocessed road environment information to obtain a trained spatiotemporal feature extraction network;

[0010] Combining the real-time perceived driving environment data with the trained feature extraction network predictions to obtain the environmental perception results and lane change decision prediction results;

[0011] The state representation is obtained by fusing the environment perception results and the lane-changing decision prediction results, and the optimal lane-changing strategy is adjusted by combining the state representation and the action value function.

[0012] Path planning is performed based on the adjusted optimal lane changing strategy, and a steering wheel angle signal and a vehicle speed control signal are generated according to the path planning results.

[0013] Furthermore, the construction process of the spatiotemporal feature extraction network includes:

[0014] Input the road environment information obtained by each sensor into the CNN network to obtain spatial features; specifically, the input data is passed through multiple convolution layers to extract feature data of different types of sensors, and the real-time video frame data is input into the first convolution layer to extract the spatial features of the image; the radar data is input into the second convolution layer to extract the radar data target distance and vehicle speed; the lidar data is input into the third convolution layer to extract the three-dimensional spatial structure and obstacle position in the point cloud data;

[0015] The spatial features and time series data are input into the RNN network to extract the comprehensive spatiotemporal features;

[0016] Output lane-changing decision features based on comprehensive spatiotemporal features.

[0017] Furthermore, the combining state representation and action value function to adjust the optimal lane changing strategy includes:

[0018] Under a given state representation, select an action a according to the current action value function Q(s,a);

[0019] After executing action a, determine the new state s' obtained after executing the action, the optimal action a' selected in the new state, and the corresponding reward r, and calculate the action value of the optimal action a' and the new state s';

[0020] By maximizing the action value function Q(s,a), the best action, i.e., the optimal lane changing strategy, is selected.

[0021] Further, a steering wheel angle signal and a vehicle speed control signal for controlling the vehicle are generated according to the path planning result, including:

[0022] Select the target point as the control target according to the path planning;

[0023] Calculate the steering angle of the vehicle tires according to the current position of the vehicle, the target point and the vehicle wheelbase, and determine the steering wheel angle limit for the steering angle of the vehicle tires;

[0024] Determine the target speed based on the recommended speed of the route planning and the current road conditions;

[0025] According to the deviation between the current vehicle speed and the target speed, the throttle or brake signal is calculated.

[0026] Furthermore, the acquired road environment information is preprocessed, including cleaning and normalizing the collected raw data.

[0027] Furthermore, when calculating the optimal path, the length of the path and the cost of the path are combined to select the optimal path, and the final path planning result is obtained based on the weighted sum of the two.

[0028] Furthermore, the Adam optimizer is used to train the constructed spatiotemporal feature extraction network.

[0029] The second purpose of the present invention is to propose an autonomous driving vehicle lane changing decision system, which can learn complex driving strategies and feature representations from a large amount of driving data. In the face of different complex situations, such as situations involving obstacles or traffic jams, the system can achieve accurate vehicle lane changing by real-time adjustment of steering wheel angle signals and vehicle speed control signals, thereby reducing various risks encountered during the lane changing process and ensuring safe and efficient driving.

[0030] A lane-changing decision system for an autonomous driving vehicle, comprising:

[0031] A preprocessing module, which is used to preprocess the acquired road environment information;

[0032] A network training module, which is used to train the constructed spatiotemporal feature extraction network based on the preprocessed road environment information to obtain a trained spatiotemporal feature extraction network;

[0033] A decision prediction module, which is used to combine the real-time perceived driving environment data and the trained feature extraction network prediction to obtain the environment perception result and the lane change decision prediction result;

[0034] A decision adjustment module is used to fuse the environmental perception results and the lane-changing decision prediction results to obtain a state representation, and adjust the optimal lane-changing strategy by combining the state representation and the action value function;

[0035] The control signal generation module is used to perform path planning based on the adjusted optimal lane changing strategy, and to generate a steering wheel angle signal and a vehicle speed control signal for controlling the vehicle according to the path planning result.

[0036] The third object of the present invention is to propose an autonomous driving vehicle, which is equipped with a variety of sensors, including lidar sensors, cameras and radars, which can obtain information including road conditions, traffic flow, vehicle position, speed, acceleration, etc. When facing a complex environment, a lane changing decision strategy is generated by fusing the acquired multi-source data. For different complex situations, such as those involving obstacles or congested traffic, accurate vehicle lane changing can be achieved by real-time adjustment of the steering wheel angle signal and the vehicle speed control signal, thereby reducing various risks encountered during the lane changing process and ensuring safe and efficient driving.

[0037] The third object of the present invention is to provide a program product.

[0038] A program product, which is a computer program product, includes a computer program, and when the computer program is executed by a processor, the steps in the lane change decision method for an autonomous driving vehicle described in the first purpose are implemented.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] The present invention proposes a lane changing decision method for an autonomous driving vehicle. When faced with a complex environment, a lane changing decision strategy is generated by fusing the acquired multi-source data. For different complex situations, such as those involving obstacles or traffic jams, accurate vehicle lane changing can be achieved by real-time adjustment of the steering wheel angle signal and the vehicle speed control signal, thereby reducing various risks encountered during the lane changing process and ensuring safe and efficient driving.

[0041] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0043] Figure 1 is a flow chart of a lane change decision method for an autonomous driving vehicle according to an embodiment of the present invention;

[0044] Figure 2 It is a structural block diagram of an automatic driving vehicle lane change decision system described in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0046] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0047] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0048] As mentioned in the background technology, the road traffic environment is extremely complex and changeable, including different road conditions, traffic flow, weather conditions, and the behavior of pedestrians and other vehicles. Traditional lane change decision methods, such as rule-based methods or simple machine learning methods, often fail to fully consider these complex factors and may perform poorly in complex scenarios.

[0049] The present invention proposes a lane changing decision method for an autonomous driving vehicle. When faced with a complex environment, a lane changing decision strategy is generated by fusing the acquired multi-source data. For different complex situations, such as those involving obstacles or traffic jams, accurate vehicle lane changing can be achieved by real-time adjustment of the steering wheel angle signal and the vehicle speed control signal, thereby reducing various risks encountered during the lane changing process and ensuring safe and efficient driving.

[0050] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0051] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0052] Figure 1 This is a flowchart of an automatic driving lane changing decision method based on deep learning according to an embodiment of the present invention.

[0053] like Figure 1 As shown, according to an embodiment of the present invention, a deep learning-based automatic driving lane change decision method includes the following steps:

[0054] S101: preprocessing the acquired road environment information;

[0055] Among them, road environment information is collected in real time through a variety of sensors installed on the autonomous driving vehicle;

[0056] Specifically, the various sensors include lidar sensors, cameras and radars, and the data obtained are road environment data collected by the corresponding sensors lidar sensors, cameras and radars; this information includes road conditions, traffic flow, vehicle position, speed, acceleration, etc.

[0057] At the same time, considering the diversity and comprehensiveness of the data, relevant data are also obtained from other sources, such as traffic monitoring cameras, satellite positioning systems, etc.

[0058] Among them, the preprocessing of the acquired road environment information includes cleaning, normalizing and other preprocessing operations on the collected raw data, removing noise, outliers and duplicate data, eliminating the dimensional differences between different sensor data, making the data have a unified scale, and labeling the data to provide accurate input and output information for subsequent model training.

[0059] S102: training the constructed spatiotemporal feature extraction network based on the preprocessed road environment information to obtain a trained spatiotemporal feature extraction network;

[0060] The specific steps include:

[0061] S201, inputting the road environment information acquired by each sensor into the CNN network to obtain spatial features, specifically including:

[0062] S2011, inputting the road environment information obtained by each sensor into the input layer of the CNN network;

[0063] S2012, extracting feature data of different types of sensors from the input data through multiple convolutional layers;

[0064] Among them, different convolution kernels are designed for different convolution layers to extract corresponding features.

[0065] Specifically, the real-time video frame data acquired by the vehicle camera is input into the first convolutional layer, and the spatial features of the image are extracted through the first convolutional layer, including the edge, texture, shape and other spatial features of the image;

[0066] The radar data is input into the second convolutional layer to extract features such as target distance and vehicle speed from the radar data;

[0067] The LiDAR data is input into the third convolutional layer to extract features such as the three-dimensional spatial structure and obstacle position in the point cloud data.

[0068] For example, in the first layer, the convolution kernel size is 3x3, the stride is 1, the padding is 1, and the number of output channels is 32.

[0069] Second layer: The convolution kernel size is 3x3, the stride is 1, the padding is 1, and the number of output channels is 64.

[0070] The third layer: The convolution kernel size is 3x3, the stride is 1, the padding is 1, and the number of output channels is 128.

[0071] S2013, performing nonlinear transformation on the extracted feature data of different types of sensors through activation functions, and downsampling through a pooling layer to reduce the spatial dimension of the data and retain important spatial features;

[0072] For example, use the maximum pooling layer with a pooling window size of 2x2 and a stride of 2. The pooling layer can reduce the size of the feature map and reduce the computational complexity.

[0073] S2014, after flattening the outputs of each convolutional layer and pooling layer, the spatial features are fused through a fully connected layer to obtain fused spatial features;

[0074] Specifically, the features extracted by the convolutional layer are fused and a preliminary feature vector is output;

[0075] The structure of the fully connected layer includes: the first layer: 512 neurons; the second layer: 256 neurons.

[0076] S2014, inputting the spatial features obtained by S2013 into the RNN network, and at the same time inputting time series data such as vehicle speed, acceleration, steering angle and the like into the RNN network;

[0077] S2015, the spatial features and time series data are input into the circulation layer after mathematical operations such as addition or concatenation and nonlinear transformation, and the comprehensive spatiotemporal features that have an important impact on the lane-changing decision are extracted;

[0078] Specifically, the recurrent layer uses recurrent units such as LSTM or GRU to capture the long-term dependencies of time series data;

[0079] S2016. Finally, the comprehensive spatiotemporal features that have an important impact on the lane-changing decision are output through the fully connected layer to output the lane-changing decision features;

[0080] Among them, the lane-changing decision features include whether to change lanes (output probability and judge through softmax function), lane-changing direction (output probability and judge through softmax function) and lane-changing timing (output a specific value or timestamp);

[0081] The output layer can set the number of output nodes according to task requirements, such as lane change decision results, steering wheel angle, etc.

[0082] S202, training the constructed spatiotemporal feature extraction network to obtain a trained spatiotemporal feature extraction network;

[0083] S2021. Selection of training algorithm;

[0084] Among them, the training algorithm selects the Adam optimizer, which usually performs better because it combines momentum and adaptive learning rate methods.

[0085] Loss function: Use cross-entropy loss (Cross-Entropy Loss);

[0086] For example, for steering wheel angle prediction, the Mean Squared Error Loss can be used.

[0087] S2022. Training process, forward propagation: send input data into the network to get output; calculate loss: calculate loss based on output and true label; back propagation: calculate gradient and update weight.

[0088] S2023. Parameter adjustment, learning rate: The initial learning rate is usually set to a small value (such as 0.001) and gradually decreased during training; Batch size: Select an appropriate batch size based on hardware resources, such as 32, 64, or 128; Number of iterations: Set the total number of iterations (Epochs), usually starting with 10 and gradually increasing until the model converges; Regularization: Use L1 or L2 regularization to prevent overfitting.

[0089] S103: inputting the real-time perceived driving environment data into the trained feature extraction network to predict the best time and path for lane changing;

[0090] The specific steps include:

[0091] S301, real-time perception driving environment data acquisition module, including driving environment data acquired by camera, radar, and laser radar;

[0092] S302, inputting the real-time perceived driving environment data into the CNN network, and judging the current perception result based on the output features, such as the vehicle's position, speed, obstacle information, etc.;

[0093] For example, it is possible to determine whether the current traffic is in a congested state; the determination of whether the current road is in a congested section is specifically as follows:

[0094]

[0095] Among them, Congestion is the congestion state; ρ is the traffic density, v is the average speed, ρ th is the traffic density threshold, v th is the average speed threshold;

[0096] S303, combining the time series data and the RNN network to obtain a lane change decision prediction result;

[0097] The RNN unit update rule is:

[0098] h t =f(Wxh x t +W hh h t-1 +b h ),

[0099] Among them, h t is the hidden state at the current moment, x t is the input at the current moment, h t-1 is the hidden state at the previous moment, W xh and W hh is the weight matrix, b h is the bias term, and f(·) is the activation function, usually ReLU or Tanh.

[0100] S104, fusing the perception result and the lane-changing decision prediction result to obtain a unified state representation s, and selecting the best lane-changing strategy by combining the state representation s and the action value function;

[0101] The specific steps include:

[0102] S401. First, the perception results (such as vehicle position, speed, obstacle information, etc.) and prediction information (such as possible traffic flow and road condition changes in the future) are integrated into a unified state representation s. This state representation s is the key input in the reinforcement learning algorithm, which contains all the information required for decision-making.

[0103] S402, action selection: Under a given state s, select an action a according to the current action value function Q(s,a), and this action a may represent changing lanes to the left lane, changing lanes to the right lane, or staying in the current lane, etc.;

[0104] S403, reward calculation: After executing action a, a reward r is calculated according to the execution result. This reward r reflects the immediate benefits obtained after executing the action (such as reducing traffic delays, improving safety, etc.).

[0105] S404, action value function update: Finally, use the update formula in reinforcement learning to update the action value function Q(s,a); select the best action by maximizing the action value function Q(s,a);

[0106]

[0107] Q(s,a)←Q(s,a)+α[R(s,a)+γmaxQ(s′,a′)-Q(s,a)],

[0108] Where α is the learning rate, γ is the discount factor, and Q(s′,a′) is the action value of the optimal action a′ and the new state s′;

[0109] This update formula usually includes the current action value Q(s,a), the new state s' obtained after executing the action, the optimal action a' selected in the new state, and the corresponding reward r. The perception results and prediction information indirectly participate in the update process of the action value function Q(s,a) by affecting the state representation s and reward r.

[0110] S105: performing path planning based on the optimal time and path for lane changing, and generating a vehicle operation control signal according to the path planning result;

[0111] The specific steps include:

[0112] S501, calculating the optimal path from the current position to the target lane;

[0113] In this embodiment, when calculating the optimal path, the length of the path and the cost of the path are combined to select, and each result can be weighted according to the actual situation, and the final path planning result is obtained based on the weighted sum of the two;

[0114] Specifically, the A* algorithm can be used to find the shortest path, and the cost of the path is evaluated according to the cost function C, which is expressed as:

[0115]

[0116] Among them, c i is the cost of each step on the path, and n is the total number of steps along the path;

[0117] S502, generating a vehicle operation control signal to control vehicle operation according to the path planning result;

[0118] In this embodiment, the control signal includes a steering wheel angle signal and a vehicle speed signal;

[0119] The specific steps include:

[0120] S5021, the path planning includes a series of waypoints, each of which contains information such as position coordinates (x, y), speed recommendations, curvature, etc. According to the current vehicle position, a target point is selected as the control target;

[0121] S5022, calculating the steering angle of the vehicle tires according to the current position of the vehicle, the target point and the vehicle wheelbase, and determining a steering wheel angle limit for the steering angle of the vehicle tires;

[0122] The formula for calculating the steering angle is:

[0123]

[0124] Among them, θ is the steering angle of the vehicle, L is the wheelbase of the vehicle, ω is the angle between the current heading of the vehicle and the direction of the target point, and L d is the distance from the vehicle to the target point.

[0125] S5023, determining a target speed according to the suggested speed of the path planning and the current road conditions (such as curvature, obstacles, etc.);

[0126] S5024, calculating the throttle or brake signal according to the deviation between the current vehicle speed and the target speed;

[0127]

[0128] Among them, u(t) is the control output of the throttle or brake, e(t) is the speed error signal between the target speed and the current speed, K p , K i , K d is the PID parameter.

[0129] Through the above steps, the autonomous driving system can adjust the lane changing strategy in time when it detects obstacles or traffic jams ahead to ensure safe and efficient driving.

[0130] Figure 2 is a structural block diagram of a lane change decision system for an autonomous driving vehicle according to an embodiment of the present invention. Figure 2 As shown, according to an embodiment of the present invention, a lane change decision system for an autonomous driving vehicle includes: a preprocessing module, a network training module, a decision prediction module, a decision adjustment module and a control signal generation module.

[0131] Among them, the division module is used to pre-process the acquired road environment information;

[0132] The network training module is used to train the constructed spatiotemporal feature extraction network based on the preprocessed road environment information to obtain a trained spatiotemporal feature extraction network;

[0133] A decision prediction module, which is used to combine the real-time perceived driving environment data and the trained feature extraction network prediction to obtain the environment perception result and the lane change decision prediction result;

[0134] A decision adjustment module is used to fuse the environmental perception results and the lane-changing decision prediction results to obtain a state representation, and adjust the optimal lane-changing strategy by combining the state representation and the action value function;

[0135] The control signal generation module is used to perform path planning based on the adjusted optimal lane changing strategy, and to generate a steering wheel angle signal and a vehicle speed control signal for controlling the vehicle according to the path planning result.

[0136] It should be noted that the specific implementation method of the autonomous driving vehicle lane changing decision system in the embodiment of the present invention is similar to the specific implementation method of the autonomous driving vehicle lane changing decision method in the embodiment of the present invention. Please refer to the description of the method part for details. In order to reduce redundancy, it will not be repeated here.

[0137] Furthermore, an embodiment of the present invention discloses an autonomous driving vehicle, which is provided with an autonomous driving vehicle lane change decision system as in the above embodiment. The vehicle is provided with a variety of sensors, including a lidar sensor, a camera and a radar, which can obtain information including road conditions, traffic flow, vehicle position, speed, acceleration, etc. When facing a complex environment, a lane change decision strategy is generated by fusing the acquired multi-source data. For different complex situations, such as those involving obstacles or congested traffic, accurate vehicle lane change can be achieved by adjusting the steering wheel angle signal and the vehicle speed control signal in real time, thereby reducing various risks encountered during the lane change process and ensuring safe and efficient driving.

[0138] In addition, other structures and functions of the vehicle according to the embodiment of the present invention are well known to ordinary technicians in the field, and are not described here to reduce redundancy.

[0139] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0140] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A lane-changing decision method for an autonomous driving vehicle, characterized in that: The steps include: Preprocessing the acquired road environment information; The constructed spatiotemporal feature extraction network is trained based on the preprocessed road environment information to obtain a trained spatiotemporal feature extraction network; Combining the real-time perceived driving environment data with the trained feature extraction network predictions to obtain the environmental perception results and lane change decision prediction results; The state representation is obtained by fusing the environment perception results and the lane-changing decision prediction results, and the optimal lane-changing strategy is adjusted by combining the state representation and the action value function. Path planning is performed based on the adjusted optimal lane changing strategy, and a steering wheel angle signal and a vehicle speed control signal are generated according to the path planning results.

2. The lane change decision method for an autonomous driving vehicle according to claim 1, characterized in that: The construction process of the spatiotemporal feature extraction network includes: Input the road environment information obtained by each sensor into the CNN network to obtain spatial features; specifically, the input data is passed through multiple convolution layers to extract feature data of different types of sensors, and the real-time video frame data is input into the first convolution layer to extract the spatial features of the image; the radar data is input into the second convolution layer to extract the radar data target distance and vehicle speed; the lidar data is input into the third convolution layer to extract the three-dimensional spatial structure and obstacle position in the point cloud data; The spatial features and time series data are input into the RNN network to extract the comprehensive spatiotemporal features; Output lane-changing decision features based on comprehensive spatiotemporal features.

3. The lane change decision method for an autonomous driving vehicle according to claim 1, characterized in that: The method of adjusting the optimal lane changing strategy by combining the state representation and the action value function includes: Under a given state representation, select an action a according to the current action value function Q(s,a); After executing action a, determine the new state s' obtained after executing the action, the optimal action a' selected in the new state, and the corresponding reward r, and calculate the action value of the optimal action a' and the new state s'; By maximizing the action value function Q(s,a), the best action, i.e., the optimal lane changing strategy, is selected.

4. The lane change decision method for an autonomous driving vehicle according to claim 1, characterized in that: Generate a steering wheel angle signal and a vehicle speed control signal to control the vehicle based on the path planning results, including: Select the target point as the control target according to the path planning; Calculate the steering angle of the vehicle tires according to the current position of the vehicle, the target point and the vehicle wheelbase, and determine the steering wheel angle limit for the steering angle of the vehicle tires; Determine the target speed based on the recommended speed of the route planning and the current road conditions; According to the deviation between the current vehicle speed and the target speed, the throttle or brake signal is calculated.

5. The lane change decision method for an autonomous driving vehicle according to claim 1, characterized in that: The acquired road environment information is preprocessed, including cleaning and normalizing the collected raw data.

6. The lane change decision method for an autonomous driving vehicle according to claim 1, characterized in that: When calculating the optimal path, the length of the path and the cost of the path are combined to make a selection, and the final path planning result is obtained based on the weighted sum of the two.

7. The lane change decision method for an autonomous driving vehicle according to claim 1, characterized in that: The Adam optimizer is used to train the constructed spatiotemporal feature extraction network.

8. An autonomous driving vehicle lane change decision system, characterized in that: include: A preprocessing module, which is used to preprocess the acquired road environment information; A network training module, which is used to train the constructed spatiotemporal feature extraction network based on the preprocessed road environment information to obtain a trained spatiotemporal feature extraction network; A decision prediction module, which is used to combine the real-time perceived driving environment data and the trained feature extraction network prediction to obtain the environment perception result and the lane change decision prediction result; A decision adjustment module is used to fuse the environmental perception results and the lane-changing decision prediction results to obtain a state representation, and adjust the optimal lane-changing strategy by combining the state representation and the action value function; The control signal generation module is used to perform path planning based on the adjusted optimal lane changing strategy, and to generate a steering wheel angle signal and a vehicle speed control signal for controlling the vehicle according to the path planning result.

9. An autonomous driving vehicle, characterized in that: An automatic driving vehicle lane changing decision system as described in claim 7 is provided.

10. A program product, the program product being a computer program product, comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps in the lane change decision method for an autonomous driving vehicle as described in any one of claims 1-7 are implemented.

Images