Rule-based setting scenario library automatic driving decision system, method and vehicle

Abstract

The application discloses a rule-based setting scene library automatic driving decision system and method and a vehicle, and belongs to the technical field of automatic driving, and comprises a sensor module, a data processing module, a decision module and an execution module.The sensor module is used for collecting environmental information around the vehicle.The data processing module is connected with the sensor module, and is used for processing and analyzing the environmental information collected by the sensor module, so as to identify an obstacle and extract feature information of the obstacle.The decision module formulates an avoidance decision according to the feature information extracted by the data processing module based on a predetermined rule.The execution module is used for controlling the vehicle to execute the avoidance decision generated by the decision module.The application can accurately identify various obstacles and their moving states in a complex urban traffic scene, rapidly judge and handle the situation of sudden appearance of the obstacles, and continuously optimize the decision rule through real-time learning and experience accumulation, so as to adapt to different urban environments and obstacle arrangements.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and more specifically, to an autonomous driving decision-making system, method, and vehicle based on a rule-based scenario library. Background Technology

[0002] With the rapid development of autonomous driving technology, how vehicles can make safe and efficient decisions in complex and ever-changing urban environments has become a critical issue that urgently needs to be addressed. Currently, the decision-making process for autonomous driving mainly relies on the real-time processing of sensor data and machine learning algorithms. This allows vehicles to recognize obstacles and traffic rules in their surroundings to a certain extent. However, when the scenario transforms into a complex situation with multiple obstacles, these conventional methods reveal shortcomings such as poor decision-making efficiency and questionable safety guarantees. For example, in urban road scenarios, vehicles may need to quickly weigh and avoid multiple pedestrians, numerous other vehicles, and static obstacles, which undoubtedly places stringent demands on the real-time performance and accuracy of the decision-making algorithm.

[0003] Currently, existing obstacle avoidance decision-making methods can be broadly categorized into two types: model-based control strategies and learning-based intelligent decision-making. While these methods can perform well in specific scenarios, they struggle to overcome several limitations when placed in dynamic environments with multiple obstacles:

[0004] First, decision-making is delayed. Given the need for comprehensive perception and analysis of complex scenarios, existing methods often inevitably cause decision-making delays, which in turn pose a threat to driving safety.

[0005] Secondly, they lack environmental adaptability. Most existing technologies rely on pre-trained models, which perform poorly when encountering new and unfamiliar obstacle distribution patterns.

[0006] Third, it consumes a lot of computing resources. Complex algorithms require a large amount of computing resources, which limits their application in real-time systems.

[0007] The existing patent, "Lane Decision-Making Method Based on Multi-Objective Decision Matrix for Autonomous Vehicles" (Patent No. CN201810383811.2), proposes a lane decision-making method based on a multi-objective decision matrix for autonomous vehicles. This method uses the left lane, the current lane, and the right lane of the autonomous vehicle as lanes to be decided. It quantitatively evaluates the distance that can be continued in each lane based on the vehicle's position and the speed limit information. Simultaneously, it collects information on the nearest obstacles to the vehicle on both sides of each lane, calculating the distances between the obstacles and the vehicle at the current time and after a set time, and compiling this information to obtain a decision matrix. The decision matrix is ​​processed through two processes: decision information weight calculation and decision analysis, enabling real-time autonomous lane decision-making for autonomous vehicles.

[0008] However, the aforementioned existing technologies have the following drawbacks: Firstly, these existing decision-making methods primarily focus on lane-specific decisions. In complex urban environments with multiple obstacles, they lack specificity for obstacle avoidance decisions, making it difficult to make quick and effective avoidance decisions. Secondly, while they consider factors such as the relative distance between obstacles in front and behind, they do not fully utilize information such as the type and movement status of obstacles, resulting in poor flexibility when dealing with dynamically changing obstacles. Thirdly, the virtual obstacle assumption problem can lead to unreasonable decisions under special circumstances, failing to accurately map real-world road conditions. Summary of the Invention

[0009] One objective of this invention is to provide a rule-based autonomous driving decision-making system with a set scenario library to address the aforementioned issues in urban environments where autonomous driving systems must cope with highly dynamic and complex traffic scenarios. Traditional perception and decision-making systems often fail to effectively integrate information from multiple sensors, resulting in insufficient obstacle recognition, state prediction, and path planning, thereby increasing the risk of collisions.

[0010] According to a first aspect of the present invention, a rule-based autonomous driving decision-making system based on a set scenario library is provided, comprising:

[0011] The sensor module is used to collect environmental information around the vehicle;

[0012] A data processing module is connected to the sensor module. The data processing module is used to process and analyze the environmental information collected by the sensor module in order to identify obstacles and extract the feature information of the obstacles.

[0013] A decision-making module is connected to the data processing module. The decision-making module formulates an avoidance decision based on predetermined rules and the feature information extracted by the data processing module.

[0014] An execution module is connected to the decision module, and the execution module controls the vehicle to execute the avoidance decision generated by the decision module.

[0015] Optionally, the sensor module includes millimeter-wave radar, lidar, and a camera, wherein the millimeter-wave radar is used to detect the speed information of the obstacle, the lidar is used to provide three-dimensional spatial information of the obstacle to help determine the accurate position and shape of the obstacle, and the camera is used to acquire image information of the obstacle.

[0016] Optionally, the data processing module uses a deep neural network model with a multi-layer convolutional neural network structure to process the images captured by the camera in real time. Specifically, it determines the position and movement status of pedestrians by detecting human feature points and contours, identifies parked vehicles based on their shape features and color information, and identifies construction areas by recognizing construction signs and fence features.

[0017] Optionally, the decision-making module has a built-in online learning module, which dynamically adjusts the decision rules through an online learning algorithm; the online learning module is equipped with an adaptive learning rate adjustment module, which automatically increases the learning rate when the vehicle enters an unfamiliar scene or when the road condition complexity is detected to increase sharply, so as to speed up the initial adjustment of the decision rules.

[0018] Optionally, the decision-making module employs a graph-based path planning algorithm, which randomly samples the space around the vehicle based on the real-time detected multiple obstacle information, and expands the path based on the position, size, and movement status of the obstacles.

[0019] Optionally, while the vehicle is driving according to the avoidance decision, it continuously senses the environmental information around the vehicle and updates the feature information of the obstacles.

[0020] Optionally, the vehicle uploads its own experience data to a cloud server and obtains experience data from other vehicles from the cloud. The experience data includes obstacle type, obstacle location, handling method, and decision result.

[0021] Optionally, the rule-based scenario library autonomous driving decision-making system also includes an evaluation module. The evaluation module has several performance evaluation indicators built-in, including safety indicators, efficiency indicators, and comfort indicators. The safety indicators include the number of collisions and the minimum distance to obstacles. The efficiency indicators include driving time and average speed. The comfort indicators include the rate of change of acceleration and the rate of change of steering angle. The performance of the autonomous driving system in multi-obstacle scenarios is evaluated and verified through these performance evaluation indicators.

[0022] According to a second aspect of the present invention, a rule-based autonomous driving decision-making method based on a set scenario library is provided, comprising:

[0023] Collect environmental information around the vehicle;

[0024] The environmental information is processed and analyzed to identify obstacles and extract their feature information;

[0025] Based on predetermined rules, an avoidance decision is made according to the feature information extracted by the data processing module.

[0026] Control the vehicle to execute the avoidance decision generated by the decision module.

[0027] According to a third aspect of the present invention, a vehicle is provided, including the above-described rule-based scenario library-based autonomous driving decision-making system.

[0028] The rule-based scenario library-based autonomous driving decision-making system, method, and vehicle disclosed herein have the following technical effects:

[0029] The decision-making module formulates optimal obstacle avoidance decisions based on rule definitions, perception technology, and learning capabilities. A rule-based scenario library provides predefined decision rules for rapid and effective obstacle avoidance in specific situations. The decision-making module incorporates an online learning module that dynamically adjusts the decision rules using online learning algorithms. This online learning module is equipped with an adaptive learning rate adjustment module, which automatically increases the learning rate when the vehicle enters an unfamiliar scenario or detects a sharp increase in road complexity, thus accelerating the initial adjustment of the decision rules. This not only significantly improves decision-making speed but also enhances the system's adaptability and robustness in complex environments by introducing flexible rule combinations. By constructing a scenario library covering common obstacle configurations and combining it with real-time sensor data, autonomous vehicles can quickly make reasonable decisions when facing complex scenarios, thereby improving safety and efficiency.

[0030] Other features and advantages of the invention will become clear from the following detailed description of exemplary embodiments of the invention with reference to the accompanying drawings. Attached Figure Description

[0031] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments of the invention and, together with their description, serve to explain the principles of the invention.

[0032] Figure 1 This is a logical diagram of the rule-based scenario library autonomous driving decision-making method provided in an embodiment of the present invention. Detailed Implementation

[0033] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention.

[0034] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use.

[0035] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0036] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0037] This invention proposes an embodiment of an autonomous driving decision-making system based on a rule-based scenario library, specifically, as follows: Figure 1 As shown, it includes:

[0038] The sensor module is used to collect environmental information around the vehicle;

[0039] A data processing module is connected to the sensor module. The data processing module is used to process and analyze the environmental information collected by the sensor module in order to identify obstacles and extract the feature information of the obstacles.

[0040] The data processing module utilizes a high-performance processor to rapidly process and analyze sensor data.

[0041] A decision-making module is connected to the data processing module. The decision-making module formulates an avoidance decision based on predetermined rules and the feature information extracted by the data processing module.

[0042] An execution module is connected to the decision module, and the execution module controls the vehicle to execute the avoidance decision generated by the decision module.

[0043] Specifically, the rule configuration in this invention involves setting multiple dynamic rules, enabling the system to make corresponding avoidance decisions based on the nature of different obstacles. For example, when encountering a pedestrian, the system automatically prioritizes pedestrian safety and avoids them according to the "pedestrian priority" rule. When the obstacle is a stationary construction area, the system evaluates the optimal detour route to ensure driving safety. This flexibility helps improve driving safety and passenger comfort.

[0044] This invention implements an independent design for the three core modules: perception, decision-making, and execution. While operating independently, each module coordinates efficiently through standardized data interfaces and communication protocols. The perception module integrates multiple sensing devices such as visual sensors, LiDAR, and ultrasonic sensors. The decision-making module is closely connected to the perception module, using real-time received accurate perception data and built-in advanced algorithms to quickly analyze complex road conditions and instantly generate optimal avoidance path plans for different scenarios. The execution module strictly follows the instructions of the decision-making module, precisely controlling every detail of the vehicle's actions through deep integration with the vehicle's acceleration, braking, and steering control systems.

[0045] In this embodiment of the invention, the sensor module includes a millimeter-wave radar, a lidar, and a camera. The millimeter-wave radar is used to detect the speed information of obstacles, the lidar is used to provide three-dimensional spatial information of the obstacles to help determine their accurate location and shape, and the camera is used to acquire image information of the obstacles. The camera, based on a visual sensor, can provide rich image information. The data detected by the millimeter-wave radar, lidar, and camera can improve the accuracy and reliability of obstacle detection, especially in complex urban environments.

[0046] In this embodiment of the invention, the data processing module uses a deep neural network model with a multi-layer convolutional neural network structure to process the images captured by the camera in real time. Specifically, it determines the position and movement status of pedestrians by detecting human feature points and contours, identifies parked vehicles based on their shape features and color information, and determines the construction area by recognizing construction signs and fence features.

[0047] In this embodiment of the invention, the decision-making module has a built-in online learning module, which dynamically adjusts the decision rules through an online learning algorithm. The online learning module is equipped with an adaptive learning rate adjustment module, which automatically increases the learning rate when the vehicle enters an unfamiliar scene or detects a sharp increase in road condition complexity, thereby accelerating the initial adjustment speed of the decision rules. This invention can make instantaneous decisions in the event of sudden obstacles, relying on a perception module to continuously update data and calculate in real time the distance between the vehicle and the obstacle, the relative speed, and the available surrounding avoidance space. Once an emergency occurs, such as an animal suddenly entering the driving lane or a vehicle illegally changing lanes, the system immediately initiates an emergency assessment procedure. If sufficient avoidance space is determined, the decision-making module quickly plans a new safe path, and the execution module seamlessly connects, smoothly adjusting the vehicle speed and direction, flexibly weaving and detouring to ensure the journey is not disturbed. If the assessment finds that the space is too cramped to avoid, the system decisively triggers an emergency braking command, and the braking system responds at full power, bringing the vehicle to a smooth stop in the shortest possible time.

[0048] In this embodiment of the invention, the decision module employs a graph-based path planning algorithm, which randomly samples the space around the vehicle based on real-time detected multiple obstacle information, and expands the path based on the position, size, and movement status of the obstacles.

[0049] Specifically, when a pedestrian is detected, the system of this invention immediately activates the core rule of "pedestrian priority". It not only actively slows down, but also accurately predicts the pedestrian's movement trajectory and reserves sufficient safety space to ensure that pedestrians can pass through without worry. If a stationary obstacle such as a construction area is encountered, the system quickly switches to detour mode. Taking into account factors such as surrounding traffic flow, road width, and intersection distribution, the system uses a complex path planning algorithm to calculate the best detour route in real time to ensure driving continuity and safety.

[0050] In this embodiment of the invention, the vehicle continuously senses environmental information around the vehicle and updates the feature information of obstacles while driving according to the avoidance decision.

[0051] In this embodiment of the invention, the vehicle uploads its own experience data to a cloud server and obtains experience data from other vehicles from the cloud. This experience data includes obstacle types, obstacle locations, handling methods, and decision results. By analyzing and learning from the experience data of other vehicles obtained from the cloud, the vehicle can accelerate its learning speed and improve the accuracy and adaptability of its decisions. For example, when a vehicle encounters a rare obstacle arrangement in a new urban area, it can obtain experience data from other vehicles in similar areas from the cloud, thereby making the correct decision more quickly.

[0052] Specifically, at each decision-making moment, the system of this invention automatically and comprehensively records key data such as obstacle type, overall environmental conditions, and final decision path to construct a decision database. Based on an online learning algorithm, the system periodically backtracks and analyzes past decision performance, rewarding successful obstacle avoidance and efficient driving cases, and penalizing collision risks and inefficient decision trajectories, thereby continuously fine-tuning and optimizing the built-in decision rules.

[0053] In this embodiment of the invention, when the sensor module detects the sudden appearance of an obstacle, the vehicle immediately activates an emergency decision-making mechanism. First, it quickly determines the type, location, speed, and direction of the obstacle using sensor information. Then, considering the vehicle's own state and the surrounding environment, it determines whether there is sufficient space to avoid the obstacle. If there is space, a new avoidance path is quickly generated based on the current path planning algorithm, and the vehicle's driving direction and speed are adjusted; if there is insufficient space, the vehicle immediately applies the brakes, comes to a safe stop, and waits for the obstacle to be removed or for a suitable driving opportunity to arise.

[0054] In this embodiment of the invention, the vehicle continuously collects data such as environmental information, obstacle characteristics, and decision results during operation. The decision rules are dynamically adjusted and optimized using an online learning algorithm. For example, when the vehicle encounters similar obstacle arrangements multiple times in a specific scenario, the online learning algorithm can adjust its decision-making strategy for that scenario based on this experience data, improving the accuracy and efficiency of handling similar situations in the future. A scenario could be a specific intersection where a large number of electric vehicles frequently cross during rush hour, and delivery trucks are often parked on the roadside. The online learning algorithm uses pattern recognition technology to identify common features of similar scenarios, such as the common driving trajectories of electric vehicles at that intersection during specific time periods and the probability of delivery trucks opening their doors to unload goods. Then, based on preset optimization objectives (such as minimizing the risk of collisions with obstacles, maximizing driving efficiency, and improving passenger comfort), the online learning algorithm adjusts the decision rules. For example, the usual decision at this intersection was to maintain a fixed distance and pass at a constant speed. However, it was found that the electric vehicles suddenly changed lanes, causing them to brake suddenly. The online learning algorithm then adjusted to slow down in advance and increase the lateral safety distance. At the same time, it collected information such as the electric vehicle's turn signal status and the rider's head rotation to predict the details of the lane change intention and change the speed control and path fine-tuning strategies.

[0055] In this embodiment of the invention, the rule-based scenario-based autonomous driving decision-making system further includes an evaluation module. This evaluation module incorporates several performance evaluation indicators, including safety, efficiency, and comfort indicators. The safety indicators include the number of collisions and the minimum distance to obstacles; the efficiency indicators include travel time and average speed; and the comfort indicators include the rate of change of acceleration and the rate of change of steering angle. These performance evaluation indicators are used to evaluate and verify the performance of the autonomous driving system in multi-obstacle scenarios. In simulation experiments, various complex urban environments and obstacle arrangements are constructed using virtual reality technology to conduct large-scale testing and optimization of the system. In field tests, representative urban roads are selected for testing, and actual data is collected and analyzed. The system parameters and algorithms are continuously adjusted and optimized to improve the system's performance and stability.

[0056] The execution module is connected to the vehicle's acceleration, braking, and steering systems via wired or wireless communication links. Wired communication links are based on automotive bus technologies such as CAN (Controller Area Network) and FlexRay. Wireless communication links utilize automotive wireless communication protocols such as Bluetooth Low Energy (BLE), Wi-Fi 6E, and 5G vehicle-to-everything (V2X) technology. The execution module's communication links not only control basic acceleration, braking, and steering functions but also coordinate with Electronic Stability Program (ESP) and adaptive suspension. During high-speed cornering, the execution module instructs the steer while simultaneously informing the ESP via the communication link to prepare for intervention, fine-tuning the braking force distribution to each wheel to prevent skidding. When encountering road surface undulations, the adaptive suspension adjusts its damping coefficient, with commands transmitted instantaneously via the link to ensure stable driving.

[0057] The present invention also provides an embodiment of a rule-based autonomous driving decision-making method based on a set scenario library, comprising:

[0058] Collect environmental information around the vehicle;

[0059] The environmental information is processed and analyzed to identify obstacles and extract their feature information;

[0060] Based on predetermined rules, an avoidance decision is made according to the feature information extracted by the data processing module.

[0061] Control the vehicle to execute the avoidance decision generated by the decision module.

[0062] It should be noted that the execution process and other contents of the above methods are based on the same concept as the system embodiments of this application. They are methods corresponding to the above-mentioned rule-based scenario library autonomous driving decision-making system. All implementation structures in the above system embodiments are applicable to the embodiments of this method. For details on their specific functions and the resulting technical effects, please refer to the system embodiments section, which will not be repeated here.

[0063] The present invention also provides an embodiment of a vehicle, including an autonomous driving decision-making system based on a rule-based scenario library. Specifically, when encountering special and complex scenarios, the vehicle of the present invention can reasonably adjust its speed, maintain a safe distance, and select an appropriate avoidance route based on stored experience and rules of similar scenarios, effectively coping with these complex and challenging traffic conditions.

[0064] The above description of the structure, features, and effects of the present invention is based on the embodiments shown in the figures. However, the above are only preferred embodiments of the present invention. It should be noted that the technical features involved in the above embodiments and their preferred methods can be reasonably combined and matched by those skilled in the art to form a variety of equivalent solutions without departing from or changing the design concept and technical effects of the present invention. Therefore, the present invention is not limited to the scope of implementation shown in the figures. Any changes made in accordance with the concept of the present invention, or modifications to equivalent embodiments, that do not exceed the spirit covered by the specification and figures, should be within the protection scope of the present invention.

Claims

1. A rule-based autonomous driving decision-making system with a set scenario library, characterized in that, include: The sensor module is used to collect environmental information around the vehicle; A data processing module is connected to the sensor module. The data processing module is used to process and analyze the environmental information collected by the sensor module in order to identify obstacles and extract the feature information of the obstacles. A decision-making module is connected to the data processing module. The decision-making module formulates an avoidance decision based on predetermined rules and the feature information extracted by the data processing module. An execution module is connected to the decision module, and the execution module controls the vehicle to execute the avoidance decision generated by the decision module; The sensor module includes a millimeter-wave radar, a lidar, and a camera. The millimeter-wave radar is used to detect the speed information of the obstacle, the lidar is used to provide the three-dimensional spatial information of the obstacle to help determine the accurate position and shape of the obstacle, and the camera is used to acquire image information of the obstacle. The decision-making module employs a graph-based path planning algorithm, which randomly samples the space around the vehicle based on real-time detected multiple obstacle information, and expands the path based on the location, size, and movement status of the obstacles. During the process of making avoidance decisions, the vehicle continuously senses the environmental information around the vehicle and updates the feature information of obstacles. The vehicle uploads its own experience data to a cloud server and obtains experience data from other vehicles from the cloud. The experience data includes obstacle type, obstacle location, handling method, and decision result. The predetermined rules are dynamic decision-making rules. The system makes corresponding avoidance decisions based on the nature of different obstacles. At each decision moment, the system automatically records the obstacle type, the overall environment, and key data of the final decision path to build a decision database. The system periodically backtracks and analyzes past decision performance, rewards successful avoidance, punishes collision risks, and dynamically optimizes the decision-making rules.

2. The rule-based autonomous driving decision-making system based on a scenario library as described in claim 1, characterized in that, The data processing module uses a deep neural network model with a multi-layer convolutional neural network structure to process the images captured by the camera in real time. Specifically, it determines the position and movement status of pedestrians by detecting human feature points and contours, identifies parked vehicles based on their shape features and color information, and identifies construction areas by recognizing construction signs and fence features.

3. The rule-based autonomous driving decision-making system based on a scenario library as described in claim 1, characterized in that, The decision-making module has a built-in online learning module, which dynamically adjusts the decision rules through an online learning algorithm. The online learning module is equipped with an adaptive learning rate adjustment module, which automatically increases the learning rate when the vehicle enters an unfamiliar scene or when the road condition complexity is detected to increase sharply, so as to speed up the initial adjustment of the decision rules.

4. The rule-based autonomous driving decision-making system based on a scenario library as described in claim 1, characterized in that, It also includes an evaluation module, which has several performance evaluation indicators built in. These indicators include safety indicators, efficiency indicators, and comfort indicators. The safety indicators include the number of collisions and the minimum distance to obstacles. The efficiency indicators include driving time and average speed. The comfort indicators include the rate of change of acceleration and the rate of change of steering angle. These performance evaluation indicators are used to evaluate and verify the performance of the autonomous driving system in multi-obstacle scenarios.

5. A rule-based autonomous driving decision-making method based on a set scenario library, characterized in that, include: Collect environmental information around the vehicle; The environmental information is processed and analyzed to identify obstacles and extract their feature information; Based on predetermined rules, avoidance decisions are made using feature information extracted by the data processing module. Control the vehicle to execute the avoidance decision generated by the decision module; The decision-making module employs a graph-based path planning algorithm, which randomly samples the space around the vehicle based on real-time detected multiple obstacle information, and expands the path based on the location, size, and movement status of the obstacles. During the process of making avoidance decisions, the vehicle continuously senses the environmental information around the vehicle and updates the feature information of obstacles. The vehicle uploads its own experience data to a cloud server and obtains experience data from other vehicles from the cloud. The experience data includes obstacle type, obstacle location, handling method, and decision result. The predetermined rules are dynamic decision-making rules. The system makes corresponding avoidance decisions based on the nature of different obstacles. At each decision moment, the system automatically and comprehensively records the obstacle type, overall environment, and key data of the final decision path to build a decision database. The system periodically backtracks and analyzes past decision performance, rewards successful avoidance, punishes collision risks, and dynamically optimizes the decision-making rules.

6. A vehicle, characterized in that, Including the rule-based scenario library autonomous driving decision-making system as described in any one of claims 1 to 4.

Images