Automatic driving decision-making method and system based on artificial intelligence

Abstract

The invention discloses an automatic driving decision-making method and system based on artificial intelligence, and relates to the technical field of data processing, and the method comprises the steps: obtaining the initial positioning and destination information of a target vehicle, and generating a target driving route; the method comprises the following steps: acquiring multi-source environment sensing data through an intelligent sensing module, and performing multivariate feature extraction to generate environment feature information; based on the environment characteristic information, a dynamic learning optimization technology is adopted to carry out real-time adjustment on the target driving route, and a driving strategy adapting to a complex traffic scene is generated; and finally, realizing automatic driving control of the target vehicle according to the real-time dynamic driving strategy. According to the method, the technical problems of incomplete environment perception, static path planning and insufficient decision real-time performance of the existing automatic driving technology in a complex traffic environment are solved, and the technical effects of comprehensive perception and real-time dynamic path planning in the complex environment are achieved through multi-source data fusion and dynamic learning optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an autonomous driving decision-making method and system based on artificial intelligence. Background Art

[0002] With the rapid development of autonomous driving technology, its application in complex traffic environments faces many challenges. For example, traditional perception systems rely on a single sensor and are difficult to handle complex and changing traffic scenarios; path planning algorithms are mostly statically preset and cannot dynamically respond to real-time traffic condition changes; decision-making systems lack efficient learning and optimization capabilities, resulting in insufficient response speed to emergencies; the accuracy of the control execution link is limited, making it difficult to achieve efficient and stable vehicle control. These problems seriously restrict the reliability and safety of autonomous driving technology in complex environments. Summary of the Invention

[0003] This application provides an autonomous driving decision-making method and system based on artificial intelligence, which is used to solve the technical problems of incomplete environmental perception, static path planning, and insufficient decision-making real-time performance existing in the existing autonomous driving technology in complex traffic environments.

[0004] In the first aspect of this application, an autonomous driving decision-making method based on artificial intelligence is provided. The method includes: obtaining the initial vehicle positioning and destination information of the target vehicle, and generating a target driving route of the target vehicle based on the initial vehicle positioning and destination information; obtaining the environmental perception data of the target vehicle through an intelligent sensing module, where the environmental perception data includes but is not limited to image data, radar data, lidar data, and GPS data; performing multi-feature extraction on the environmental perception data to generate environmental feature information; dynamically learning and optimizing the target driving route based on the environmental feature information to generate a real-time dynamic driving strategy; and performing autonomous driving control of the target vehicle according to the real-time dynamic driving strategy.

[0005] In the second aspect of the present application, an artificial intelligence-based autonomous driving decision-making system is provided. The system includes: a driving route generation module, which is used to obtain the initial vehicle positioning and destination information of the target vehicle, and generate the target driving route of the target vehicle based on the initial vehicle positioning and destination information; an environment perception module, which is used to obtain the environment perception data of the target vehicle through an intelligent sensing module, and the environment perception data includes but is not limited to image data, radar data, lidar data, and GPS data; an environment feature extraction module, which is used to perform multi-source feature extraction on the environment perception data to generate environment feature information; a dynamic driving strategy generation module, which is used to dynamically learn and optimize the target driving route based on the environment feature information to generate a real-time dynamic driving strategy; and an autonomous driving control module, which is used to perform autonomous driving control of the target vehicle according to the real-time dynamic driving strategy.

[0006] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0007] The artificial intelligence-based autonomous driving decision-making method and system provided in the present application relate to the technical field of data processing. By obtaining the initial vehicle positioning and destination to generate a driving route, using multi-source sensors to collect environment data and extract features, dynamically optimizing the route based on the environment features, and generating a real-time driving strategy to achieve autonomous driving control, it solves the technical problems of incomplete environment perception, static path planning, and insufficient decision-making real-time performance in existing autonomous driving technologies in complex traffic environments, and realizes the technical effect of comprehensive perception and real-time dynamic path planning in complex environments through multi-source data fusion and dynamic learning optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0009] Figure 1 It is a schematic flowchart of an artificial intelligence-based autonomous driving decision-making method provided by an embodiment of the present application;

[0010] Figure 2 It is a schematic structural diagram of an artificial intelligence-based autonomous driving decision-making system provided by an embodiment of the present application.

[0011] Description of the attached drawing reference numerals: Driving route generation module 11, environmental perception module 12, environmental feature extraction module 13, dynamic driving strategy generation module 14, and automatic driving control module 15. Detailed implementation manners

[0012] The present application provides an artificial intelligence-based automatic driving decision-making method and system, which are used to solve the technical problems of incomplete environmental perception, static path planning, and insufficient decision-making real-time performance in existing automatic driving technologies in complex traffic environments.

[0013] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0014] It should be noted that the terms "first", "second", etc. in the specification and the above-mentioned attached drawings of the present application are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server including a series of steps or units does not necessarily need to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0015] Embodiment 1, as Figure 1 shown, the present application provides an artificial intelligence-based automatic driving decision-making method, and the method includes:

[0016] P10: Obtain the initial vehicle positioning and destination information of the target vehicle, and generate the target driving route of the target vehicle based on the initial vehicle positioning and destination information.

[0017] Furthermore, step P10 of the embodiment of the present application further includes:

[0018] P11: Simulate the driving path from the initial vehicle positioning to the destination through map navigation data in combination with real-time traffic information, and generate multiple candidate routes; P12: Select the optimal route from the multiple candidate routes as the target driving route based on a preset optimization goal.

[0019] It should be understood that the initial location and destination information of the target vehicle are obtained, and the target driving route of the target vehicle is generated based on this information. This process not only requires precise positioning technology but also needs to comprehensively consider the real-time traffic conditions and optimization objectives to ensure the rationality and efficiency of the route.

[0020] First, by combining map navigation data with real-time traffic information, the driving path from the initial vehicle location to the destination is simulated, and multiple candidate routes are generated. Specifically, high-precision digital map data is first called. These maps not only contain the geometric shapes of roads but also cover detailed information such as road types, lane numbers, traffic signals, speed limit signs, etc. At the same time, real-time traffic data is accessed. This data may come from the vehicle-to-everything (V2X) of the traffic management department or the vehicle's own sensors to sense the surrounding traffic flow. By analyzing this data, it is judged which sections may be congested and which sections have higher traffic efficiency. Based on this information, path planning algorithms (such as the Dijkstra algorithm) are used to simulate multiple possible paths from the starting point to the end point. These paths will consider different road choices, traffic signal waiting times, and possible traffic congestion situations, thus generating multiple candidate routes. This process provides a rich selection basis for subsequent route optimization.

[0021] Next, based on the preset optimization objectives, the optimal route is selected from multiple candidate routes as the target driving route. The optimization objective is the key decision factor for path planning and can be set according to the user's preferences and the specific needs of the vehicle. For example, the optimization objective can be the shortest time, the shortest distance, the safest route, or the lowest energy consumption, etc. According to these optimization objectives, combined with the preset weights (for example, the time weight is the highest, followed by safety and energy consumption), each candidate route is comprehensively evaluated. For example, if the optimization objective is to select the route with the shortest time consumption, the system will calculate the fastest arrival path by combining the real-time traffic flow and speed limit information; if the optimization objective is to select a safer route, the system will avoid accident-prone sections or construction areas. Finally, an optimal route is selected from multiple candidate routes as the target driving route and transmitted to the subsequent modules of the autonomous driving system to guide the vehicle's dynamic driving strategy and control. This path planning method that comprehensively considers multiple factors provides more intelligent and efficient basic decision support for autonomous driving vehicles.

[0022] P20: Through the intelligent sensing module, the environmental perception data of the target vehicle is obtained, and the environmental perception data includes but is not limited to image data, radar data, lidar data, and GPS data.

[0023] Furthermore, step P20 of the embodiment of the present application further includes:

[0024] P21: Deploy an intelligent sensing module, where the intelligent sensing module includes an image sensor, a radar sensor, and a lidar sensor; P22: Collect obstacle image data around the target vehicle through the image sensor; P23: Obtain obstacle position data around the target vehicle through the radar sensor; P24: Collect three-dimensional point cloud data around the target vehicle through the lidar sensor; P25: Construct environmental perception data of the target vehicle from the obstacle image data, obstacle position data, and three-dimensional point cloud data.

[0025] Optionally, by deploying an intelligent sensing module, obtain environmental perception data around the target vehicle, which includes image data, radar data, lidar data, and GPS data. These multi-source data provide basic support for subsequent feature extraction and decision-making.

[0026] First, a complete set of intelligent sensing modules needs to be deployed to achieve all-round perception of the vehicle's surrounding environment. The module includes an image sensor, a radar sensor, and a lidar sensor. The image sensor is used to capture visual information around the vehicle and can identify traffic signs, lane lines, pedestrians, etc.; the radar sensor obtains the position and speed information of obstacles around the vehicle by emitting millimeter-wave signals; the lidar sensor generates three-dimensional point cloud data of the vehicle's surrounding environment through laser scanning. The collaborative work of these sensors provides the vehicle with rich environmental perception capabilities.

[0027] Among them, the image sensor is responsible for collecting obstacle image data around the target vehicle. These image data are the basis for the vision perception of the autonomous driving system. The image sensor captures the surrounding scene of the vehicle through a high-resolution camera and can identify traffic signs, lane lines, pedestrians, and other potential obstacles. For example, through deep learning algorithms, the system can analyze the image data, quickly identify the content of traffic signs, and judge whether the vehicle's driving position deviates from the lane based on the lane lines. The high resolution and rich color information of the image data make it one of the important sources of perception data in the autonomous driving system.

[0028] The radar sensor is responsible for obtaining obstacle position data around the target vehicle. The radar sensor can accurately measure the distance and speed of obstacles by emitting millimeter-wave signals and receiving reflected signals. These data are crucial for the autonomous driving system. Especially in high-speed driving scenarios, the radar sensor can monitor the speed changes of the vehicle ahead in real time and provide decision-making basis for the vehicle's automatic following and emergency braking. For example, when the radar sensor detects that the vehicle ahead suddenly decelerates, the system can quickly react and adjust the vehicle's driving speed to avoid collisions. The long-distance detection ability and adaptability to bad weather of the radar sensor make it an indispensable part of the autonomous driving system.

[0029] The lidar sensor is responsible for collecting the three-dimensional point cloud data around the target vehicle. This data can generate a high-precision three-dimensional model of the vehicle's surrounding environment, providing richer spatial information for the autonomous driving system. By emitting laser beams and measuring the time difference of the reflected light, the lidar can accurately measure the distance and shape of obstacles. For example, the lidar can detect potholes on the road, the contours of obstacles, and details in other complex environments. These three-dimensional point cloud data can not only help the vehicle identify the positions of obstacles, but also generate an environmental map around the vehicle through point cloud processing algorithms, providing support for path planning and obstacle avoidance decisions.

[0030] Fuse the collected obstacle image data, obstacle position data, and three-dimensional point cloud data to construct the environmental perception data of the target vehicle. This process involves multi-sensor data fusion technology, aiming to integrate different types of perception data into a complete environmental model. For example, the system can identify the types of obstacles through image data, obtain the speed and distance information of obstacles through radar data, and then combine the three-dimensional point cloud data of the lidar to generate a comprehensive environmental model including the positions, shapes, and motion states of obstacles. This multi-modal data fusion not only improves the accuracy of environmental perception but also enhances the robustness of the system, enabling it to make more reliable decisions in complex and changing driving scenarios.

[0031] P30: Extract multi-features from the environmental perception data to generate environmental feature information.

[0032] Furthermore, step P30 of the embodiment of the present application further includes:

[0033] P31: Extract target type features from the obstacle image data to generate multiple obstacle types; P32: Process the obstacle position data to extract the physical displacement features of the obstacles; P33: Segment and classify the three-dimensional point cloud data to identify the road and obtain spatial obstacle features; P34: Fuse the multiple obstacle types, the physical displacement features, and the spatial obstacle features to generate comprehensive environmental feature information.

[0034] Specifically, through multi-feature extraction technology, key information is extracted from complex perception data to generate environmental feature information for subsequent decision-making. This process involves the comprehensive processing of image data, position data, and three-dimensional point cloud data to ensure that the autonomous driving system can accurately understand the surrounding environment and make reasonable decisions.

[0035] First, using the obstacle image data collected by the image sensor, target type features are extracted. This process uses deep learning algorithms (such as convolutional neural networks, CNNs) to identify and classify objects in the image, thereby generating various obstacle types. For example, pedestrians, vehicles, traffic signs, lane lines, and other static or dynamic obstacles are identified. By extracting target type features, the nature of surrounding objects can be quickly determined, providing an important basis for subsequent decision-making. For example, when a pedestrian is identified ahead, the system can adjust the vehicle speed in advance or issue a warning to ensure safety.

[0036] Next, the obstacle position data obtained by the radar sensor is processed to extract the physical displacement features of the obstacle. These features include information such as the speed, acceleration, and movement direction of the obstacle. By analyzing these physical displacement features, the system can predict the future position and movement trajectory of the obstacle. For example, if it is detected that the vehicle ahead is decelerating, the system can adjust the vehicle speed in advance to avoid emergency braking. The extraction of physical displacement features provides real-time information about the dynamic environment for the autonomous driving system, enabling it to make quick responses in complex traffic scenarios.

[0037] Furthermore, the three-dimensional point cloud data collected by the lidar sensor is segmented and classified. Through the point cloud segmentation algorithm, the system can distinguish between the road, obstacles, and background regions, and further classify the obstacles. For example, the system can identify potholes on the road, the shape and size of obstacles, and the position of the road boundary through the point cloud data. This information is integrated into spatial obstacle features, providing high-precision three-dimensional environment information for the autonomous driving system. The extraction of spatial obstacle features enables the system to make more accurate obstacle avoidance decisions in complex road conditions (such as construction areas or narrow roads).

[0038] Finally, the obstacle types extracted from the image data, the physical displacement features extracted from the radar data, and the spatial obstacle features extracted from the three-dimensional point cloud data are fused. This fusion process uses multimodal data fusion algorithms to integrate different types of data into a comprehensive environmental feature information. For example, the system can combine the obstacle types identified from the image data and the movement trajectories predicted from the radar data to generate a complete obstacle information model. At the same time, by fusing the spatial obstacle features in the three-dimensional point cloud data, the system can more comprehensively understand the surrounding environment. This comprehensive environmental feature information provides comprehensive and accurate support for the subsequent decision-making of the autonomous driving system, enabling it to make safer and more efficient decisions in complex and changing driving scenarios. This process not only improves the system's perception ability of complex environments but also enhances its adaptability and reliability in dynamic scenarios.

[0039] P40: Based on the environmental feature information, dynamically learn and optimize the target driving route to generate a real-time dynamic driving strategy.

[0040] Further, step P40 of the embodiment of the present application further includes:

[0041] P41: According to the environmental feature information, conduct a multi-dimensional effect evaluation on the target driving route, including evaluating the safety, feasibility, and driving efficiency of the current driving route, and generating a multi-dimensional effect evaluation coefficient; P42: Based on the multi-dimensional effect evaluation coefficient, generate an optimization vector of driving parameters; P43: Based on the optimization vector of driving parameters, optimize the target driving route to generate a real-time dynamic driving strategy.

[0042] It should be understood that the target driving route is dynamically optimized according to the extracted environmental feature information, and a real-time dynamic driving strategy is generated. This process ensures that the autonomous driving system can make safe, efficient, and adaptable decisions in a complex and changing environment through multi-dimensional effect evaluation and parameter optimization.

[0043] First, use the extracted environmental feature information to conduct a multi-dimensional effect evaluation on the target driving route. This evaluation covers three core dimensions: safety, feasibility, and driving efficiency, to generate a multi-dimensional effect evaluation coefficient. The safety evaluation analyzes the type, position, and movement trajectory of obstacles to determine whether there are potential risks on the route; the feasibility evaluation considers whether the route conforms to traffic rules and the road passing conditions; the driving efficiency evaluation analyzes factors such as the passing time, distance, and energy consumption of the route. By comprehensively evaluating these dimensions, the system generates a quantitative effect evaluation coefficient to provide a basis for subsequent optimization.

[0044] Next, generate an optimization vector of driving parameters according to the multi-dimensional effect evaluation coefficient. This vector contains key parameters for adjusting the driving route, such as speed, acceleration, steering angle, and obstacle avoidance strategy. For example, if the route safety is low, the system will reduce the speed to increase the reaction time; if the route efficiency is high and there is no risk, the speed will be appropriately increased. By optimizing these parameters, the system can provide specific guidance for dynamic adjustment of the autonomous driving vehicle to ensure that the vehicle makes optimal decisions in a complex environment.

[0045] Finally, optimize the target driving route according to the driving parameter optimization vector and generate a real-time dynamic driving strategy. This process adjusts the key parameters of the driving route to ensure an optimal balance among vehicle safety, feasibility, and efficiency. For example, if road congestion is detected ahead, the system will adjust the route and select a smoother alternative path; if an obstacle is detected, the system will adjust the steering angle and speed to avoid the obstacle safely. The finally generated real-time dynamic driving strategy not only considers the current environmental characteristics but also adapts to the changing road conditions through dynamic learning and optimization to ensure the safety and efficiency of autonomous driving.

[0046] Further, step P43 of the embodiment of the present application further includes:

[0047] P43-1: Extract a speed optimization vector based on the driving parameter optimization vector and generate a vehicle speed control strategy; P43-2: Extract a lane optimization vector according to the driving parameter optimization vector and generate a lane change strategy; P43-3: Extract an obstacle avoidance optimization vector according to the driving parameter optimization vector and generate a vehicle obstacle avoidance strategy; P43-4: Integrate the vehicle speed control strategy, lane change strategy, and vehicle obstacle avoidance strategy, correlate the target driving route, and generate the real-time dynamic driving strategy.

[0048] In a possible embodiment of the present application, the target driving route is optimized according to the driving parameter optimization vector to generate a real-time dynamic driving strategy. This process ensures that the vehicle makes safe, efficient, and adaptable decisions in a complex environment by extracting and integrating multiple optimization vectors.

[0049] First, extract a speed optimization vector based on the driving parameter optimization vector and generate a vehicle speed control strategy. The speed optimization vector includes parameters such as target speed, acceleration, and deceleration. The system calculates the optimal speed control strategy based on environmental characteristic information (such as road speed limits, speeds of vehicles ahead, traffic signal states) and safety and efficiency evaluation results. For example, reduce the speed in a congested section to maintain a safe distance, or increase the speed in a smooth section to improve driving efficiency. The speed control strategy is implemented through the vehicle's power system (such as the throttle and brakes).

[0050] Next, extract a lane optimization vector according to the driving parameter optimization vector and generate a lane change strategy. The lane optimization vector includes parameters such as target lane, lane change timing, and lane change trajectory. The system determines whether a lane change is needed and the specific way of the lane change based on environmental characteristic information (such as lane lines, positions of surrounding vehicles, traffic rules) and feasibility evaluation results. For example, choose to overtake when encountering a slow vehicle, or change lanes in advance when approaching an exit. The lane change strategy is implemented through the vehicle's steering system.

[0051] Furthermore, based on the driving parameter optimization vector, an obstacle avoidance optimization vector is extracted to generate a vehicle obstacle avoidance strategy. The obstacle avoidance optimization vector includes parameters such as the obstacle avoidance path, the obstacle avoidance timing, and the obstacle avoidance action. The system plans the obstacle avoidance path and actions according to the environmental feature information (such as the position of obstacles, the movement trajectory, and the road geometric structure) and the safety assessment results. For example, decelerate or detour when detecting a pedestrian ahead, or adjust the driving trajectory when encountering a static obstacle. The obstacle avoidance strategy is realized by comprehensively controlling the speed and steering system of the vehicle.

[0052] Next, the vehicle speed control strategy, the lane change strategy, and the vehicle obstacle avoidance strategy are fused to associate the target driving route and generate a real-time dynamic driving strategy. The system integrates the speed control, lane change, and obstacle avoidance strategies into a unified driving strategy through a multi-strategy fusion algorithm (such as weighted fusion, priority scheduling) to ensure the coordination and consistency among various strategies. For example, adjust the speed and the obstacle avoidance path simultaneously during a lane change to avoid collisions with other vehicles. The generated real-time dynamic driving strategy is executed through the vehicle control system to achieve safe, efficient, and comfortable autonomous driving.

[0053] Through this fusion method, the system can generate a comprehensive real-time dynamic driving strategy to ensure the vehicle travels safely and efficiently in a complex and changeable environment.

[0054] P50: Perform autonomous driving control of the target vehicle according to the real-time dynamic driving strategy.

[0055] Furthermore, step P50 of the embodiment of the present application further includes:

[0056] P51: Convert the real-time dynamic driving strategy into real-time vehicle control instructions and send them to the vehicle actuator; P52: Through the vehicle actuator, split the vehicle control instructions into a multi-step control execution sequence and execute the autonomous driving of the target vehicle according to the multi-step control execution sequence.

[0057] Specifically, it is a key step to convert the real-time dynamic driving strategy into specific actions. This process converts the strategy into vehicle control instructions and realizes the autonomous driving function through the vehicle actuator to ensure that the vehicle can travel safely and efficiently according to the optimized strategy.

[0058] First, the real-time dynamic driving strategy is converted into specific vehicle control instructions and sent to the vehicle's actuators. This process involves converting the abstract decisions in the strategy (such as speed adjustment, lane change, obstacle avoidance, etc.) into executable instructions, such as acceleration, deceleration, steering angle adjustment, etc. These instructions are transmitted to the actuators, such as motors, brakes, and steering systems, through the vehicle's electronic control unit (ECU). For example, if the real-time dynamic driving strategy requires the vehicle to decelerate to avoid an obstacle ahead, the system will generate a corresponding deceleration instruction and send it to the vehicle's braking system to ensure that the vehicle can decelerate in time.

[0059] Next, after receiving the control instructions, the vehicle actuators split them into a series of multi-step control execution sequences. This is because the physical execution process of the vehicle needs to be completed step by step. For example, a lane change may require adjusting the steering angle first, then the speed, and finally completing the lane change. By splitting the control instructions into multi-step execution sequences, the vehicle can execute the autonomous driving task more smoothly and safely. For example, when the vehicle needs to change lanes from the current lane to the left lane, the actuator will first turn on the turn signal, then gradually adjust the steering angle while maintaining an appropriate speed, and finally complete the lane change operation. This step-by-step execution method not only improves the reliability of autonomous driving but also enhances the adaptability and flexibility of the system.

[0060] Through the above steps, the system can convert the real-time dynamic driving strategy into specific vehicle control instructions and realize the autonomous driving function through the vehicle actuators. This process ensures that the vehicle can drive safely and efficiently according to the optimized strategy, and at the same time provides reliable technical support for the practical application of the autonomous driving system.

[0061] In summary, the embodiments of the present application have at least the following technical effects:

[0062] In the present application, by obtaining the initial positioning and destination information of the target vehicle, generating a target driving route, collecting multi-source environmental perception data for multi-feature extraction to generate environmental feature information, based on the environmental feature information, using dynamic learning optimization technology to adjust the target driving route in real time, generating a driving strategy adapted to complex traffic scenarios, and finally realizing the autonomous driving control of the target vehicle according to the real-time dynamic driving strategy.

[0063] It achieves the technical effect of realizing comprehensive perception and real-time dynamic path planning in complex environments through multi-source data fusion and dynamic learning optimization.

[0064] Embodiment 2, based on the same inventive concept as the artificial intelligence-based autonomous driving decision-making method in the foregoing embodiment, such as Figure 2As shown in the figure, the present application provides an autonomous driving decision-making system based on artificial intelligence. The system and method embodiments in the embodiments of the present application are based on the same inventive concept. Among them, the system includes:

[0065] A driving route generation module 11, which is configured to obtain the initial vehicle positioning and destination information of the target vehicle, and generate a target driving route of the target vehicle based on the initial vehicle positioning and destination information.

[0066] An environment perception module 12, which is configured to obtain environment perception data of the target vehicle through an intelligent sensing module, and the environment perception data includes but is not limited to image data, radar data, lidar data, and GPS data.

[0067] An environment feature extraction module 13, which is configured to perform multi-feature extraction on the environment perception data to generate environment feature information.

[0068] A dynamic driving strategy generation module 14, which is configured to perform dynamic learning and optimization on the target driving route based on the environment feature information to generate a real-time dynamic driving strategy.

[0069] An autonomous driving control module 15, which is configured to perform autonomous driving control of the target vehicle according to the real-time dynamic driving strategy.

[0070] Further, the driving route generation module 11 is further configured to perform the following steps:

[0071] Simulate the driving path from the initial vehicle positioning to the destination through map navigation data in combination with real-time traffic information to generate multiple candidate routes; select the optimal route from the multiple candidate routes as the target driving route based on a preset optimization target.

[0072] Further, the environment perception module 12 is further configured to perform the following steps:

[0073] Deploy an intelligent sensing module, where the intelligent sensing module includes an image sensor, a radar sensor, and a lidar sensor; collect obstacle image data around the target vehicle through the image sensor; obtain obstacle position data around the target vehicle through the radar sensor; collect three-dimensional point cloud data around the target vehicle through the lidar sensor; construct the environment perception data of the target vehicle from the obstacle image data, obstacle position data, and three-dimensional point cloud data.

[0074] Further, the environment feature extraction module 13 is further configured to perform the following steps:

[0075] Extract target type features from the obstacle image data to generate multiple obstacle types; process the obstacle position data to extract the physical displacement features of the obstacles; segment and classify the 3D point cloud data to identify the road and obtain the spatial obstacle features; fuse the multiple obstacle types, the physical displacement features, and the spatial obstacle features to generate comprehensive environmental feature information.

[0076] Further, the dynamic driving strategy generation module 14 is further configured to perform the following steps:

[0077] According to the environmental feature information, perform a multi-dimensional effect evaluation on the target driving route, including evaluating the safety, feasibility, and driving efficiency of the current driving route, to generate a multi-dimensional effect evaluation coefficient; based on the multi-dimensional effect evaluation coefficient, generate an optimized vector of driving parameters; based on the optimized vector of driving parameters, optimize the target driving route to generate a real-time dynamic driving strategy.

[0078] Further, the dynamic driving strategy generation module 14 is further configured to perform the following steps:

[0079] Based on the optimized vector of driving parameters, extract the optimized vector of speed, and generate a vehicle speed control strategy; according to the optimized vector of driving parameters, extract the optimized vector of lane, and generate a lane change strategy; according to the optimized vector of driving parameters, extract the optimized vector of obstacle avoidance, and generate a vehicle obstacle avoidance strategy; fuse the vehicle speed control strategy, the lane change strategy, and the vehicle obstacle avoidance strategy, and associate them with the target driving route to generate the real-time dynamic driving strategy.

[0080] Further, the autonomous driving control module 15 is further configured to perform the following steps:

[0081] Convert the real-time dynamic driving strategy into real-time vehicle control instructions and send them to the vehicle actuator; through the vehicle actuator, split the vehicle control instructions into a multi-step control execution sequence, and perform autonomous driving of the target vehicle according to the multi-step control execution sequence.

[0082] It should be noted that the above sequence of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above description of specific embodiments of this specification is provided. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multi-tasking and parallel processing are also possible or may be advantageous.

[0083] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0084] This specification and the drawings are merely exemplary illustrations of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An automatic driving decision-making method based on artificial intelligence, characterized in that: The method comprises: Acquiring initial vehicle positioning and destination information of the target vehicle, and generating a target driving route of the target vehicle based on the initial vehicle positioning and destination information; Acquire environmental perception data of the target vehicle through the intelligent sensor module, the environmental perception data including but not limited to image data, radar data, lidar data and GPS data; Performing multi-dimensional feature extraction on the environmental perception data to generate environmental feature information; Based on the environmental feature information, dynamically learning and optimizing the target driving route to generate a real-time dynamic driving strategy; According to the real-time dynamic driving strategy, automatic driving control of the target vehicle is performed.

2. The automatic driving decision-making method based on artificial intelligence according to claim 1, characterized in that: Based on the initial vehicle positioning and destination information, a target driving route of the target vehicle is generated, including: By combining map navigation data with real-time traffic information, the driving path from the initial vehicle location to the destination is simulated to generate multiple candidate routes; Based on a preset optimization target, an optimal route is selected from the plurality of candidate routes as a target driving route.

3. The automatic driving decision-making method based on artificial intelligence according to claim 1, characterized in that: Through the intelligent sensor module, the environmental perception data of the target vehicle is obtained, including: Deploy intelligent sensor modules, including image sensors, radar sensors and lidar sensors; Collect obstacle image data around the target vehicle through an image sensor; Obtain obstacle location data around the target vehicle through radar sensors; Collect 3D point cloud data around the target vehicle through the LiDAR sensor; The obstacle image data, obstacle position data and three-dimensional point cloud data are used to construct environmental perception data of the target vehicle.

4. The automatic driving decision-making method based on artificial intelligence according to claim 3, characterized in that: Performing multivariate feature extraction on the environmental perception data to generate environmental feature information includes: Extracting target type features according to the obstacle image data to generate multiple obstacle types; Processing the obstacle position data to extract physical displacement characteristics of the obstacle; Segmenting and classifying the three-dimensional point cloud data to identify the road and obtain spatial obstacle features; The multiple obstacle types, the physical displacement characteristics, and the spatial obstacle characteristics are integrated to generate comprehensive environmental characteristic information.

5. The automatic driving decision-making method based on artificial intelligence according to claim 1, characterized in that: Based on the environmental feature information, the target driving route is dynamically learned and optimized to generate a real-time dynamic driving strategy, including: Based on the environmental feature information, a multi-dimensional effect evaluation is performed on the target driving route, including evaluating the safety, feasibility and driving efficiency of the current driving route, and generating a multi-dimensional effect evaluation coefficient; Based on the multi-dimensional effect evaluation coefficient, generating a driving parameter optimization vector; The target driving route is optimized based on the driving parameter optimization vector to generate a real-time dynamic driving strategy.

6. The automatic driving decision-making method based on artificial intelligence according to claim 5, characterized in that: The target driving route is optimized based on the driving parameter optimization vector to generate a real-time dynamic driving strategy, including: Based on the driving parameter optimization vector, extracting a speed optimization vector and generating a vehicle speed control strategy; Extracting a lane optimization vector based on the driving parameter optimization vector and generating a lane change strategy; Extracting an obstacle avoidance optimization vector based on the driving parameter optimization vector and generating a vehicle obstacle avoidance strategy; The vehicle speed control strategy, lane change strategy and vehicle obstacle avoidance strategy are integrated, the target driving route is associated, and the real-time dynamic driving strategy is generated.

7. The automatic driving decision-making method based on artificial intelligence according to claim 1, characterized in that: According to the real-time dynamic driving strategy, automatic driving control of the target vehicle is performed, including: Convert real-time dynamic driving strategies into real-time vehicle control commands and send them to vehicle actuators; The vehicle control instruction is split into a multi-step control execution sequence through a vehicle actuator, and the automatic driving of the target vehicle is executed according to the multi-step control execution sequence.

8. The autonomous driving decision-making system based on artificial intelligence is characterized by: The system comprises: A driving route generation module, the driving route generation module is used to obtain an initial vehicle location and destination information of a target vehicle, and generate a target driving route of the target vehicle based on the initial vehicle location and destination information; An environmental perception module, which is used to obtain environmental perception data of the target vehicle through an intelligent sensor module, wherein the environmental perception data includes but is not limited to image data, radar data, lidar data and GPS data; An environmental feature extraction module, the environmental feature extraction module is used to extract multivariate features from the environmental perception data to generate environmental feature information; A dynamic driving strategy generation module, the dynamic driving strategy generation module is used to dynamically learn and optimize the target driving route based on the environmental feature information to generate a real-time dynamic driving strategy; An automatic driving control module is used to perform automatic driving control of a target vehicle according to the real-time dynamic driving strategy.

Images