Control method and device for automobile, storage medium and terminal

Abstract

The invention discloses a control method and device for an automobile, a storage medium and a terminal. The control method comprises the following steps that first state data of the automobile and second state data of surrounding automobiles are obtained; based on the first state data and the second state data, a control strategy of the automobile is determined, and the control strategy comprises a following strategy and a lane changing strategy; and controlling the movement of the automobile based on the control strategy. According to the control method, the movement of the automobiles can be controlled in the multi-automobile cooperative formation.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and more particularly to control methods, devices, storage media, and terminals for automobiles. Background Technology

[0002] In existing multi-vehicle cooperative platooning control technologies, numerous studies and patents have proposed various control methods and systems in an attempt to improve the stability, response speed, and adaptability of the platoons. However, these technologies still have significant shortcomings in complex environments.

[0003] 1. Limitations of Centralized and Distributed Control: Lu Ruoyu's proposed method for autonomous driving platooning, combining centralized and distributed approaches, achieves distributed computational burden by selecting a central vehicle. However, in complex environments, it still relies on the central vehicle for trajectory planning. This means that if the central vehicle fails or information is delayed, the system's stability will be affected. Furthermore, while distributed control can reduce the computational burden, it lacks sufficient redundancy mechanisms and struggles to effectively cope with the demands of dynamic environments. Especially in the event of a central vehicle failure, the system will face performance degradation or lag.

[0004] 2. Lack of adaptability to complex road environments: Deng Guochen proposed a platooning control method based on the navigation and following approach for four-wheel drive inspection vehicles and line-following unmanned vehicles. However, traditional navigation and following strategies are prone to accumulated deviations when forming platoons, especially in dynamic road environments (such as encountering sudden obstacles or changes in traffic density), making it difficult to flexibly adjust the formation. Shang Guangtao proposed an improved adsorption behavior scheme to address this problem of platooning dispersion. However, this scheme may still result in platooning disorder or ineffective obstacle avoidance when the road environment changes in a complex manner (such as sharp turns or narrowing lanes), limiting its wide applicability in practice.

[0005] 3. Inflexible Control Strategy Switching Mechanism: In multi-vehicle platooning control, existing technologies generally lack multi-level control strategy switching mechanisms. While Xu Liwei's pilot-following method achieves platoon stability through vehicle dynamics and multi-agent control algorithms, it lacks a flexible multi-level strategy adjustment mechanism during platoon state transitions (such as formation, disbandment, or lane changing), leading to instability during formation changes. Furthermore, traditional control algorithms typically only operate under specific conditions and cannot switch in real-time, making it difficult to ensure rapid response and stability of vehicles during complex formation changes.

[0006] 4. Limitations of Sensor Data Acquisition and Insufficient Control Precision: In Chen Bo's collaborative control method for unmanned mobile target vehicles, the synchronization of speed and angle of multiple target vehicles was achieved through multi-sensor data fusion and intelligent control algorithms. However, due to the limitations of sensor data, especially in multi-vehicle platooning, the control precision of each vehicle is difficult to guarantee. In traditional systems, sensors cannot cover all neighboring vehicles, making it impossible to obtain accurate information about neighboring vehicles. This results in insufficient precision in the relative position control of vehicles at high speeds or in dense platooning, increasing the safety risks of platooning, especially when the distance between vehicles is small, which may lead to collisions and other dangers.

[0007] 5. Limitations of Specific Application Scenarios: Existing multi-vehicle platooning control systems are often designed for specific scenarios, lacking cross-scenario adaptability. For example, Shang Guangtao proposed a platooning control strategy based on a virtual structure method and tested it on a smart miniature vehicle, but this method often shows insufficient adaptability in other application scenarios (such as airports, mines, and other open spaces). Most systems perform well in specific scenarios, but when applied to a wider range of real-world roads or various application environments, existing platooning control systems struggle to accommodate the different environmental characteristics and vehicle requirements of each scenario, thus limiting their applicability.

[0008] In summary, while existing multi-vehicle cooperative platooning control systems have achieved certain results in specific applications and conditions, they still face several challenges in complex environments. These challenges include: reliance on a central node for centralized control, insufficient adaptability to complex dynamic environments, inflexible control strategy switching, insufficient control accuracy due to limited sensor data, and limitations in application scenarios. These issues restrict the efficiency, stability, and widespread application of autonomous vehicle platooning systems, failing to meet diverse scenario requirements.

[0009] Therefore, how to control the movement of vehicles in multi-vehicle platooning has become an urgent problem to be solved. Summary of the Invention

[0010] In view of this, the main objective of the present invention is to provide a control method, device, storage medium and terminal for automobiles.

[0011] To achieve the above objectives, the technical solution of the present invention is implemented as follows: a control method for a car, comprising the following steps: acquiring first state data of the car and second state data of surrounding cars; determining a control strategy for the car based on the first and second state data, the control strategy including a following strategy and a lane-changing strategy; and controlling the movement of the car based on the control strategy.

[0012] As an improvement to this embodiment of the invention, the "controlling the motion of the vehicle based on the control strategy" specifically includes: when the control strategy is a following strategy, performing the following operations: establishing a first dynamic model of the vehicle in the longitudinal direction, in which the longitudinal acceleration of the vehicle is... The change is The longitudinal position of the vehicle The relationship with speed is ,in, Let be the longitudinal velocity of the car, t be time, m be the mass of the car, and k be a constant; the objective function of the first dynamic model is to minimize the value of J. Where N is a natural number, For the car and the car in front at time The actual distance at that time For the car and the car in front at time Safe distance at time For the car at time The longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at time; the constraint condition of the first dynamic model is: the first preset threshold ≤ acceleration. ≤Second preset threshold, 0≤speed The threshold values ​​are: ≤ third preset threshold, <0, and >0. Within the first time period, the objective function in the first dynamic model is minimized based on the position and speed of the car and the position and speed of the car in front, thereby obtaining the acceleration of the car. Within the second time period, the movement of the car is controlled based on the acceleration. Both the first and second time periods start from the current time, and the length of the first time period is > the length of the second time period.

[0013] As an improvement to this embodiment of the invention, the step of "minimizing the objective function in the first dynamic model based on the position and speed of the car and the position and speed of the car in front" specifically includes: minimizing the objective function in the first dynamic model using gradient descent or sequential quadratic programming based on the position and speed of the car and the position and speed of the car in front.

[0014] As an improvement to this embodiment of the invention, the "controlling the motion of the vehicle based on the control strategy" specifically includes: when the control strategy is a following strategy, performing the following operations: establishing a first dynamic model of the vehicle in the longitudinal direction and a second dynamic model in the lateral direction, wherein in the second dynamic model, the lateral acceleration of the vehicle is... The change is The position of the car in the lateral direction The relationship with speed is in, Let be the lateral velocity of the car, t be time, and m be the mass of the car; the objective function of the second dynamic model is to minimize the value of J. Where N is a natural number, For the car at time The lateral position at time, y_target is the lateral position of the target lane. For the car at time longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at that time; the constraint condition of the second dynamic model is: the fourth preset threshold ≤ acceleration. ≤ Fifth preset threshold, 0 ≤ speed The threshold value is ≤6, and the fourth, fifth, and sixth preset threshold values ​​are all greater than zero. During the first time period, the objective functions in the first and second dynamic models are minimized to obtain the longitudinal acceleration and lateral acceleration of the vehicle. During the second time period, the movement of the vehicle is controlled based on the acceleration. The first and second time periods both start from the current time, and the length of the first time period is greater than the length of the second time period.

[0015] This invention also provides a control device for a car, comprising the following modules: a data acquisition module for acquiring first state data of the car and second state data of surrounding cars; a strategy generation module for determining a control strategy for the car based on the first and second state data, the control strategy including a following strategy and a lane-changing strategy; and a control module for controlling the movement of the car based on the control strategy.

[0016] As an improvement to this embodiment of the invention, the control module is further configured to: when the control strategy is a following strategy, perform the following operations: establish a first dynamic model of the vehicle in the longitudinal direction, in which the longitudinal acceleration of the vehicle is... The change is The longitudinal position of the vehicle The relationship with speed is ,in, Let be the longitudinal velocity of the car, t be time, m be the mass of the car, and k be a constant; the objective function of the first dynamic model is to minimize the value of J. Where N is a natural number, For the car and the car in front at time The actual distance at that time For the car and the car in front at time Safe distance at time For the car at time The longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at time; the constraint condition of the first dynamic model is: the first preset threshold ≤ acceleration. ≤Second preset threshold, 0≤speed The threshold values ​​are: ≤ third preset threshold, <0, and >0. Within the first time period, the objective function in the first dynamic model is minimized based on the position and speed of the car and the position and speed of the car in front, thereby obtaining the acceleration of the car. Within the second time period, the movement of the car is controlled based on the acceleration. Both the first and second time periods start from the current time, and the length of the first time period is > the length of the second time period.

[0017] As an improvement to this embodiment of the invention, the control module is further configured to: based on the position and speed of the vehicle and the position and speed of the vehicle in front, minimize the objective function in the first dynamic model using gradient descent or sequential quadratic programming.

[0018] As an improvement to this embodiment of the invention, the control module is further configured to: when the control strategy is a following strategy, perform the following operations: establish a first dynamic model of the vehicle in the longitudinal direction and a second dynamic model in the lateral direction, wherein in the second dynamic model, the lateral acceleration of the vehicle is... The change is The position of the car in the lateral direction The relationship with speed is in, Let be the lateral velocity of the car, t be time, and m be the mass of the car; the objective function of the second dynamic model is to minimize the value of J. Where N is a natural number, For the car at time The lateral position at time, y_target is the lateral position of the target lane. For the car at time longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at that time; the constraint condition of the second dynamic model is: the fourth preset threshold ≤ acceleration. ≤ Fifth preset threshold, 0 ≤ speed The threshold value is ≤6, and the fourth, fifth, and sixth preset threshold values ​​are all greater than zero. During the first time period, the objective functions in the first and second dynamic models are minimized to obtain the longitudinal acceleration and lateral acceleration of the vehicle. During the second time period, the movement of the vehicle is controlled based on the acceleration. The first and second time periods both start from the current time, and the length of the first time period is greater than the length of the second time period.

[0019] This invention also provides a terminal, including: one or more processors; and a memory storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the control method described above.

[0020] This invention also provides a storage medium storing a computer program, which, when executed by a processor, implements the control method described above.

[0021] The control method, device, storage medium, and terminal for automobiles provided in this invention have the following advantages: This invention discloses a control method, device, storage medium, and terminal for automobiles. The control method includes the following steps: acquiring first state data of the automobile and second state data of surrounding automobiles; determining a control strategy for the automobile based on the first and second state data, the control strategy including a following strategy and a lane-changing strategy; and controlling the movement of the automobile based on the control strategy. This control method can control the movement of automobiles in multi-vehicle cooperative formations. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a control method for an automobile provided in an embodiment of the present invention. Figure 2 A distributed controller structure for individual human-vehicle vehicles; Figure 3 A roadmap for a distributed hybrid control system for coordinated convoy driving. Detailed Implementation

[0023] The present invention will now be described in detail with reference to the embodiments shown in the accompanying drawings. However, these embodiments do not limit the present invention, and any structural, methodological, or functional modifications made by those skilled in the art based on these embodiments are included within the scope of protection of the present invention.

[0024] The following description and accompanying drawings fully illustrate specific embodiments described herein to enable those skilled in the art to practice them. Some embodiments may include or substitute parts and features of other embodiments. The scope of the embodiments herein includes the entire scope of the claims and all available equivalents thereof. Throughout this document, the terms “first,” “second,” etc., are used only to distinguish one element from another without requiring or implying any actual relationship or order between the elements. Indeed, a first element can also be referred to as a second element, and vice versa. Furthermore, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a structure, apparatus, or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a structure, apparatus, or device. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the structure, apparatus, or device that includes said element. The various embodiments described herein are presented in a progressive manner, with each embodiment focusing on its differences from other embodiments; similar or identical parts between embodiments can be referred to interchangeably.

[0025] The terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer" used in this document to indicate orientation or positional relationships are based on the orientation or positional relationships shown in the accompanying drawings and are used only for the convenience of describing this document and simplifying the description. They do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting the invention. In the description herein, unless otherwise specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to mechanical or electrical connections, or internal connections between two elements, or direct connections or indirect connections through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0026] Embodiment 1 of the present invention provides a control method for automobiles, such as... Figure 1 , Figure 2 and Figure 3 As shown, it includes the following steps: Step 101: Obtain the first state data of the vehicle and the second state data of the surrounding vehicles; The first state data specifically includes the vehicle's own state data and environmental perception data, including: 1) Using LiDAR to collect data such as the 3D coordinates, distance, and direction of movement of surrounding obstacles, the vehicle can construct a 3D model of the surrounding environment, accurately locate the position and movement trend of obstacles, and provide crucial information for path planning and obstacle avoidance. For example, when a stationary obstacle is detected ahead, its size and shape can be determined based on its 3D coordinates, and a suitable detour path can be planned; if the obstacle is moving, its possible trajectory can be predicted based on its direction of movement, and an avoidance decision can be made in advance.

[0027] 2) Using data such as lane lines, traffic signs, and the outlines and colors of nearby vehicles obtained by cameras, lane line recognition helps vehicles stay within their lanes and improves driving safety; traffic sign recognition enables vehicles to comply with traffic rules, such as speed limits and no-entry signs, and avoid violations; and the outlines and colors of nearby vehicles help vehicles distinguish between different types and states of vehicles, such as ambulances and police cars, so that appropriate yielding measures can be taken in a timely manner.

[0028] 3) Real-time detection of speed, relative distance, and acceleration of nearby vehicles using millimeter-wave radar is crucial for maintaining a safe following distance and preventing rear-end collisions. Vehicles can adjust their speed in real time based on the speed and relative distance of the vehicle in front, ensuring safe following. Furthermore, by analyzing changes in the acceleration of nearby vehicles, it's possible to predict whether they will accelerate to overtake or decelerate and brake, allowing for proactive responses.

[0029] 4) The absolute position (latitude and longitude), heading angle, and acceleration data of the vehicle are obtained using GPS (Global Positioning System) / IMU (Inertial Measurement Unit). The absolute position (latitude and longitude) information of the vehicle is used to determine the specific location of the vehicle on a global scale, and combined with map data, it can realize the navigation function; the heading angle data helps the vehicle determine its own driving direction and ensures that the vehicle travels along the set route; the acceleration data can be used to analyze the vehicle's motion state, such as acceleration, deceleration, or constant speed driving, and provide a reference for the control of the vehicle's power system and the adjustment of the braking system.

[0030] 5) Real-time vehicle speed and steering angle data obtained from wheel speed sensors are used to monitor the vehicle's real-time speed and steering angle. Real-time speed information is the basic data for speed control and following control, while steering angle information is used for functions such as steering control and lane keeping assist. For example, based on the current speed and steering angle, the vehicle can calculate the turning radius and adjust the speed in advance to ensure a smooth and safe turning process.

[0031] Next, these data need to be preprocessed, specifically including: 1) Filtering: The data acquired by the sensors is filtered to remove noise and interference signals. For example, the Kalman filter algorithm can be used to filter data from LiDAR and millimeter-wave radar, utilizing their predictive and updated state estimates to reduce the impact of measurement noise and improve data accuracy. For camera image data, median filtering, Gaussian filtering, and other methods can be used to remove salt-and-pepper noise and Gaussian noise from the image, enhancing image clarity.

[0032] 2) Coordinate transformation: Unifying data collected by different sensors into the same coordinate system for data fusion. For example, transforming the 3D coordinate data of LiDAR to the vehicle's body coordinate system, or converting the 2D pixel coordinates in camera images to 3D world coordinates, to achieve coordinate alignment between sensor data.

[0033] 3) Data fusion, specifically probabilistic fusion methods such as Kalman filtering and its derivative algorithms (extended Kalman filtering, unscented Kalman filtering, etc.). Taking vehicle target tracking as an example, the position and velocity information of obstacles detected by lidar are fused with the position and velocity information of the same target measured by millimeter-wave radar. By establishing a state-space model and an observation model of the target, the Kalman filtering algorithm is used to fuse and estimate the multi-sensor data, resulting in a more accurate target state estimate, including position, velocity, acceleration, and other information, thereby improving the accuracy and reliability of target tracking.

[0034] Feature-level fusion methods extract features from data collected by different sensors and then fuse these features. Taking lane detection as an example, on the one hand, camera images are used to detect lane edges and fit curves, extracting the shape and position features of the lane lines; on the other hand, the absolute position of the lane lines is corrected and supplemented by combining the vehicle's GPS position and heading angle information. By fusing these two types of feature information, the position and direction of the lane lines can be determined more accurately, providing a more reliable basis for lane-keeping assistance.

[0035] 4) Data association and target recognition Data association, in a multi-sensor environment, determines whether measurement data from different sensors belong to the same target. For example, by calculating the similarity between the target's position and velocity detected by different sensors, or by matching the target's appearance features (such as color and shape), measurement data of the same target can be associated to form comprehensive observation information of the target.

[0036] Target recognition utilizes deep learning algorithms (such as convolutional neural networks) to analyze fused data and identify target types (e.g., cars, pedestrians, bicycles), brands, and models. For car targets, it can further identify features such as license plate numbers and car colors, providing richer information for traffic management and intelligent driving applications.

[0037] 5) Information updates and state estimation Based on the fused data and target recognition results, information about the vehicle's surrounding environment is updated in real time, including the location, speed, and type of obstacles, as well as road conditions (such as lane line positions and traffic sign content). Simultaneously, by combining the vehicle's own dynamics model and control algorithms, the vehicle's motion state is estimated and predicted, providing the vehicle's decision-making layer with accurate environmental perception and self-state information to enable reasonable driving decisions, such as acceleration, deceleration, steering, and obstacle avoidance.

[0038] Second-state data can be obtained from surrounding vehicles via V2X (Vehicle To Everything), specifically including: 1) Real-time data on the location, speed, acceleration, and direction of travel of surrounding vehicles; 2) Data such as the control strategy status of surrounding vehicles (e.g., cruise, follow, lane change, etc.); 3) Data such as the overall route planning and formation objectives of the convoy.

[0039] Next, the second-state data can be preprocessed, specifically as follows: 1) Time synchronization: Since the system times of different cars may differ, it is necessary to synchronize the acquired data of surrounding cars. This can be achieved by using methods such as Network Time Protocol (NTP) or Global Positioning System (GPS) timestamps to unify the data of all cars under the same time base, ensuring data consistency and accuracy.

[0040] 2) Data format conversion: Different vehicles may use different data formats and encoding methods, which need to be converted into a unified format for subsequent processing. For example, location data should be unified into latitude and longitude coordinates or Cartesian coordinates, and speed and acceleration data should be converted into the same units and dimensions.

[0041] 3) Data cleaning: Remove obviously erroneous or abnormal data points, such as speed or acceleration values ​​that exceed reasonable ranges, or location information that is clearly inconsistent with the status of surrounding vehicles. Data cleaning can be performed by setting thresholds and ensuring the consistency and continuity of the detected data.

[0042] Then, the second-state data can be fused and correlated. 1) Data fusion: This involves fusing data collected by the device's own sensors with data shared by surrounding vehicles to obtain more comprehensive and accurate environmental perception information. For example, obstacle information detected by the device's own LiDAR can be fused with obstacle information reported by surrounding vehicles, and methods such as weighted averaging and Kalman filtering can be used to obtain more accurate estimates of obstacle position, speed, and size.

[0043] 2) Data association: Determine whether the target in the data shared by surrounding vehicles and the corresponding data from your own sensors is the same object. This can be done by comparing the similarity of information such as position, speed, and direction of travel, or by using the target's unique identifier (such as a vehicle ID). For successfully associated data, integrate it into a unified target description, including multi-dimensional information from your own sensors and surrounding vehicles.

[0044] 3) Information Updates and State Estimation, including: Information Updates: Based on the fused data, real-time updates are made to the vehicle's surrounding environment, including dynamic information such as the position, speed, acceleration, and direction of travel of surrounding vehicles, as well as global information such as the overall path planning and formation goals of the convoy. This updated information provides the vehicle's decision-making layer with the latest environmental perception basis. State Estimation: Combining the vehicle's own dynamics model and control algorithms, the vehicle's motion state is estimated and predicted. For example, based on the current speed, acceleration, and direction of travel, the changes in the vehicle's position and speed over a future period are predicted; simultaneously, considering the motion state and possible driving intentions of surrounding vehicles, their impact on the vehicle's driving is estimated, providing support for subsequent collaborative decision-making.

[0045] Next, the first and second data integration methods need to be improved. Specifically, multi-sensor data fusion algorithms (such as Kalman filtering algorithms) and communication protocol parsing techniques are used to align the sensor data and communication data in time and space, eliminating redundancy and noise.

[0046] The Kalman filter algorithm includes the following steps: 1. State prediction: Based on the state estimate from the previous moment and the system's dynamic model (such as a car's dynamics model), predict the current state value. For example, for the car's position and speed, the current position and speed can be predicted based on the previous position, speed, and acceleration. 2. Measurement update: Compare the actual measured values ​​collected by the sensors with the predicted values ​​and calculate the residual (the difference between the measured and predicted values). Then, based on the Kalman gain matrix, correct the predicted values ​​to obtain a more accurate state estimate. The calculation of the Kalman gain matrix needs to consider the covariance matrix of system noise and measurement noise. 3. Covariance update: Update the covariance matrix of the state estimation error to reflect the uncertainty of the estimate. By continuously iterating through the state prediction, measurement update, and covariance update steps, the Kalman filter can gradually approximate the true value and improve the accuracy of the data.

[0047] Spatiotemporal alignment uses the system time of the vehicle itself as a reference, performing timestamp transformation on data shared by neighboring vehicles. For control strategy state data, methods such as linear interpolation or polynomial fitting can be used to transform the control strategy state values ​​of neighboring vehicles at different timestamps to the same timestamp as the vehicle itself, achieving temporal alignment. Next, the data from different vehicles are unified to the same spatial coordinate system. For example, the position information of neighboring vehicles can be transformed from their own coordinate system to a local coordinate system centered on their own vehicle, or to a global coordinate system (such as the WGS-84 coordinate system). For control strategy states, their relative position and orientation in space can be determined by combining the vehicle's driving direction and position information, thus achieving spatial alignment.

[0048] Redundancy and noise elimination: When multiple sensors measure the same physical quantity, if the difference between the measurements is less than a certain threshold (e.g., the difference between the distance measurements of the same obstacle by lidar and millimeter-wave radar is less than 0.1 meters), the data is considered redundant. The average value or the sensor data with higher confidence can be used as the final result. If a measurement significantly deviates from its historical value or is inconsistent with other related measurements (e.g., a sudden spike in a car's acceleration far exceeding the normal range at a certain moment), the data is considered potentially noisy. Noisy data can be identified and removed by setting reasonable thresholds, using statistical methods (such as standard deviation), or using machine learning-based anomaly detection algorithms.

[0049] High-precision obstacle maps are generated by fusing LiDAR and camera data.

[0050] Coordinate transformation converts the 3D point cloud data acquired by the LiDAR from the LiDAR coordinate system to the vehicle coordinate system, and at the same time converts the pixel coordinates in the 2D image data captured by the camera to 2D plane coordinates in the vehicle coordinate system.

[0051] Feature extraction and matching involves extracting feature points such as edges and corners of obstacles from LiDAR point cloud data and extracting corresponding visual features (such as color edges and texture features) from camera image data. Feature matching algorithms (such as nearest neighbor matching and SIFT feature matching) are then used to match feature points in the LiDAR data with feature points in the camera images, establishing a correlation between the two.

[0052] Data fusion and map generation: Based on the matching results, LiDAR point cloud data and camera image data are fused. LiDAR data provides precise location and shape information of obstacles, while camera data provides appearance information such as texture and color, generating a high-precision obstacle map containing the location, shape, size, and appearance features of obstacles. The obstacle map can be stored and updated using representations such as occupancy grid maps or object-based maps.

[0053] GPS data is combined with the location information of nearby vehicles to calculate relative distances and formation deviations.

[0054] Coordinate transformation and unification converts the absolute position (latitude and longitude) of the vehicle provided by GPS data into planar coordinates in a Cartesian coordinate system, while also converting the position information reported by neighboring vehicles to the same coordinate system. Methods such as Mercator projection or UTM (Universal transverse mercartor grid System) projection can be used for coordinate transformation.

[0055] Relative distance calculation involves using the Euclidean distance formula to calculate the relative distance between the vehicle and nearby vehicles based on their planar coordinates. For example, the vehicle's coordinates are... The coordinates of the nearest car are Then the relative distance .

[0056] Formation deviation calculation involves determining the target position coordinates of each vehicle in the ideal formation based on the overall convoy path planning and formation objectives. The actual position coordinates are compared with the target position coordinates to calculate the deviation values ​​of each vehicle in the lateral (perpendicular to the direction of travel) and longitudinal (direction of travel) directions. For example, the lateral deviation can be expressed as the difference between the actual and target positions on the lateral coordinate axis, and the longitudinal deviation as the difference between the two on the longitudinal coordinate axis. By continuously calculating and updating relative distances and formation deviations, feedback information is provided for the coordinated control of the vehicles, enabling the maintenance and adjustment of the formation.

[0057] Step 102: Based on the first and second state data, determine the control strategy of the vehicle, the control strategy including: a following strategy and a lane-changing strategy; The protocol information flow generation process involves pre-defining fleet coordination rules, including priority allocation between vehicles, communication protocols, and task assignment rules, and encoding these rules into data formats that can be transmitted between vehicles, such as JSON and XML. For example, priority rules for vehicles under different driving states are defined; when a vehicle is in an emergency braking state, its priority is higher, and other vehicles should give way. Then, based on the current driving task and coordination strategy, corresponding task instructions are generated, such as speed adjustment instructions, lane change instructions, and formation adjustment instructions. These instructions are encapsulated according to a pre-defined communication protocol, with necessary header information added (such as sender ID, receiver ID, message type, timestamp, etc.) to form a complete protocol data packet. Finally, the encapsulated protocol data packet is sent to other vehicles via V2X communication technology, awaiting confirmation from the receiver. If no confirmation is received within a certain time, the packet is retransmitted or other appropriate measures are taken according to the communication protocol to ensure reliable transmission of protocol information.

[0058] The protocol information stream processing receives protocol data packets sent by other vehicles. It parses the header information and task instruction content of the data packets according to a preset communication protocol, extracting key information such as the sender ID, message type, and task instruction. Then, it verifies the legality of the received protocol information, checking whether the sender is a legitimate member of the fleet and whether the message conforms to the coordination rules. Simultaneously, it determines the priority of the protocol information according to predefined priority rules to determine the processing order. For example, emergency obstacle avoidance instructions have higher priority than general speed adjustment instructions. Next, based on the task instructions in the protocol information, it executes corresponding operations, such as adjusting vehicle speed or performing lane changes. During execution, it provides feedback to the sender, informing them whether the instruction was successfully executed and the progress, so that the sender can understand the situation and make appropriate adjustments. If the instruction cannot be executed for some reason, it promptly sends an error message to the sender and takes corresponding remedial measures according to the coordination rules, such as replanning the driving route or requesting assistance from other vehicles.

[0059] The event information stream triggers a state machine (such as a finite state machine) to make policy decisions. For example, when the event information stream contains "lane congestion ahead", the strategy switches to "split strategy" and splits the convoy into sub-convoys; when the protocol information stream contains "lead car lane change instruction", the strategy switches to "lane change strategy" and adjusts the trajectory of the following car.

[0060] The vehicle control system incorporates a dedicated event monitoring module that receives real-time event information streams from the sensor fusion module and the communication module. These streams contain various event types, such as obstacle detection events, lane-change requests from nearby vehicles, and fleet member changes. The monitoring module performs initial classification of these events for targeted processing later. Simultaneously, the event information stream serves as input to the hybrid automaton, triggering state transitions and strategy switching.

[0061] The rule base is initialized and mapped to states. A set of rules is predefined, based on a large amount of traffic scenario data, vehicle dynamics characteristics, and traffic rules, and corresponds to the discrete states of the hybrid automaton. For example, "If the current state is cruising, and no obstacle is detected ahead, and no lane change request has been received, then maintain the cruising state"; "If the current state is following, and the speed of the vehicle in front decreases below a certain threshold, and there is a lane available for switching to on the side, then consider a lane change strategy." Each rule is associated with a specific discrete state and state transition condition of the hybrid automaton.

[0062] The rules are expressed in terms of state transitions. Each rule is expressed using an "if-then" logical structure. The condition part can contain multiple sub-conditions, which are combined using logical operators (AND, OR, NOT). For example, in the rule "If the event is 'A platoonable car is detected' and the current state is cruising, then switch to the combined strategy," "A platoonable car is detected" and "The current state is cruising" are two sub-conditions. They are combined together using a logical AND operation to trigger the transition from the cruising state to the combined state.

[0063] Fuzzy sets are defined to describe state fuzziness. For factors that are difficult to quantify precisely but influence decision-making, fuzzy sets are defined. For example, the relative distance between cars can be classified into fuzzy sets such as "very close," "relatively close," "moderate," "relatively far," and "very far"; relative speed can be classified into fuzzy sets such as "rapidly approaching," "slowly approaching," "at the same speed," "slowly moving away," and "rapidly moving away." These fuzzy sets are used to describe the uncertainty of cars in different states.

[0064] Membership function determination and state matching degree calculation involve assigning a membership function to each fuzzy set to describe the degree of membership of different actual values ​​to the fuzzy set. For example, when the relative distance is 5 meters, the membership degree to the "closer" fuzzy set might be 0.7, while the membership degree to the "moderate" fuzzy set might be 0.3. Membership functions can be triangular, trapezoidal, or Gaussian functions, etc. In hybrid automata, the membership function is used to calculate the matching degree between the current state and each fuzzy set, thereby determining the fuzzy characteristics of the state.

[0065] The reasoning process and state transition decision-making take current event information and vehicle state information as input and match them with rules in the rule base. First, an exact match is performed, where the input information completely satisfies the condition part of the rule. If no exact match is found, fuzzy logic is used for fuzzy matching. For example, if the current relative distance is 6 meters and the relative speed is -3 m / s (indicating the car in front is decelerating), the membership degree of the rule to each fuzzy set is calculated using a membership function, and then the rule with the highest matching degree is found. When multiple rules match successfully, conflict resolution is performed based on rule priority and matching degree. Rules with higher priority are executed first; if priorities are the same, the rule with the highest matching degree is selected. The matching degree is determined by calculating the similarity between the input information and the rule condition part; the higher the similarity, the higher the matching degree. Based on the matching result, a decision is made on whether to trigger the hybrid automaton's state transition.

[0066] Step 103: Based on the control strategy, control the movement of the car.

[0067] Fleet coordination rules are defined in advance. For example, under normal driving conditions, the safe distance between vehicles is a fixed value (e.g., 2 meters) plus the distance traveled by multiplying the vehicle's speed by a safe time interval (e.g., 1 second). That is, safe distance = vehicle speed × safe time interval + fixed distance. Simultaneously, the communication frequency of vehicles in the convoy is specified, such as sending their own status information every 0.1 seconds, including position, speed, and acceleration.

[0068] Encoding and transmission involve encoding these coordination rules into a data format, such as JSON. For example, the rule portion could be represented as "safety-distance-formula":"speed×1+2", "communication-frequency":"10Hz". These rules are then sent as part of the protocol's information stream to other vehicles in the fleet via TCP / IP or DDS protocols.

[0069] Priority allocation is based on the vehicle's task type and real-time status. For example, vehicles performing emergency tasks (such as transporting emergency medical supplies) have the highest priority, set at 100; vehicles performing normal transportation tasks have a priority of 50. Furthermore, when a vehicle malfunctions, its priority increases according to the severity of the malfunction; for example, a minor malfunction increases to 60, and a serious malfunction increases to 90.

[0070] For transmission and updates, priority information is included in the protocol data packets, such as adding a "priority":"50" field to JSON data. Vehicles exchange priority information periodically (e.g., every second) to stay informed about changes in each other's priorities.

[0071] Command generation: Task commands are generated by the fleet management unit (which can be a pre-designated lead car or an independent management device). For example, when the fleet needs to accelerate through a green wave section, the management unit generates an acceleration command, including the target speed (e.g., 80 km / h) and the acceleration time (e.g., reaching the target speed within the next 10 seconds).

[0072] Encapsulation and transmission involve encapsulating task instructions within protocol data packets, using specific identifiers to distinguish different types of instructions. For example, an acceleration instruction could be represented as "command":"accelerate", "target-speed":"80", and "time-duration":"10". This is then transmitted to designated vehicles or the entire fleet via a communication protocol.

[0073] State definition In cruise control, the car travels at a set speed, maintaining a safe distance from other vehicles. In this state, speed control is primarily based on the car's own speed setting and a simple following control algorithm, such as adjusting its own speed according to a formula based on the speed of the vehicle ahead and the safe distance.

[0074] In following mode, the car closely follows the vehicle in front, adjusting its own speed and position in real time based on the speed and position of the vehicle in front. In this mode, the car adjusts its speed and acceleration more frequently to maintain a close following distance.

[0075] In lane-changing mode, a car performs a lane-changing operation, which includes judging the timing of the lane change and adjusting the car's lateral position. In this mode, the car needs to consider the distance and speed of vehicles to the side, as well as the traffic conditions in the target lane.

[0076] In the splitting phase, the convoy is divided into sub-convoys, and each car needs to redefine its driving goals and routes. Cars must consider coordination with other cars in the sub-convoys and how to safely break away from the original convoy.

[0077] The state transition condition, from cruise to follow, is triggered when the event information stream detects a cooperative vehicle entering a safe distance range ahead, and the following conditions are met (such as a speed difference within a certain range). For example, when the vehicle's speed is 60 km / h, and a vehicle ahead is detected to be traveling at 55-65 km / h, and the distance is within the safe distance range, the state switches from cruise to follow.

[0078] The transition from following to lane changing is triggered when a lane-changing instruction is received from the lead vehicle in the protocol information stream, or when a vehicle is detected cutting in from the side in the event information stream, potentially leading to an excessively close following distance. For example, when a lane-changing instruction is received or when a vehicle is detected to be less than a safe distance threshold (e.g., 1.5 meters), the state switches from following to lane changing.

[0079] From cruise / follow mode to split mode, when the event information stream detects that the congestion level of the lane ahead exceeds a certain threshold (e.g., lane speed is below 20 km / h and congestion length exceeds 500 meters), a state transition is triggered. At this time, the convoy begins to split into sub-formations. Cars determine their own sub-formation according to preset splitting rules (e.g., grouping by car number parity) and switch to split mode.

[0080] State transition algorithm steps: Initialize; all cars are initially in cruise mode.

[0081] Event listener continuously monitors event information streams and protocol information streams.

[0082] Conditional judgment: When an event or protocol information is detected, a judgment is made based on the above state transition conditions.

[0083] State updates: If the state transition conditions are met, the vehicle's current state is updated, and the corresponding state entry operation is executed. For example, when switching from cruise mode to follow mode, the parameters of the follow control algorithm are initialized, such as the initial value of the safe distance and the follow speed adjustment coefficient.

[0084] The process is repeated in a loop. In each new state, the corresponding state control logic is followed to perform the operation, while the system continues to listen for events and protocol information in preparation for the next state transition judgment.

[0085] Here, a combination of rule-based decision trees and fuzzy logic is used. For example, if the event is "detected as a platoonable vehicle" and the current state is cruising, the strategy is switched to the combined strategy; if the protocol information stream contains "emergency obstacle avoidance command", the strategy is switched to the split strategy.

[0086] The control objectives and constraints are: (1) Cruise strategy, the objective is to maintain vehicle speed, and the constraint is to center the lane; (2) Follow strategy, the objective is to maintain a safe distance from the vehicle in front, and the constraint is acceleration limit; (3) Lane change strategy, the objective is to complete lane change, and the constraint is lateral acceleration threshold.

[0087] After generating the execution command, the accelerator / brake command of the car is controlled based on the execution command, and the desired steering angle is converted into a steering motor control signal.

[0088] Discrete states are associated with continuous dynamics and control objectives and constraints. Discrete states represent different cooperative strategies, each corresponding to a specific continuous dynamic model and control objectives and constraints. For example, in cruise mode, the car's continuous dynamic model mainly considers maintaining a constant speed, with its equations of motion being dv / dt=a (acceleration) and dx / dt=v (velocity). The control objective is to maintain vehicle speed, and the constraint is lane centering. In following mode, the car's equations of motion need to consider maintaining the distance to the vehicle in front, such as an acceleration adjustment model based on a safe distance formula. The control objective is to maintain a safe distance from the vehicle in front, and the constraint is acceleration limitation.

[0089] The state transition triggering and control target update mechanism involves a hybrid automaton triggering a state transition when the event information flow and protocol information flow meet the switching conditions. For example, upon receiving a "navigator lane change instruction" event, the system switches from the current state to the lane change strategy state. During the switch, the vehicle's control target and constraints are updated according to the cooperative strategy corresponding to the target state. For instance, when switching from cruise mode to lane change mode, the control target changes from maintaining vehicle speed and lane centering to completing lane changes and smoothly controlling speed, and the constraints change from lane centering constraints to lateral acceleration threshold constraints and safe distance constraints from vehicles on the side.

[0090] Suppose that car A is executing a cruise control strategy at a speed of 80 km / h. At this time, the event monitoring module detects two events: first, car B in the adjacent lane on the left sends a lane change request, wanting to merge into the lane where car A is located; second, car C 100 meters ahead stops due to a malfunction and occupies part of the lane.

[0091] Event response and rule matching: Car A's control system first processes these two events. Regarding Car B's lane-change request, based on the rule in the rule base, "If a lane-change request is received from a neighboring vehicle and the current state is cruise, determine whether the lane-change acceptance conditions are met," it begins to evaluate whether to accept the request. The evaluation conditions include checking whether the expected distance after the lane change meets the safe distance requirement and the vehicle's own speed adjustment range, among others.

[0092] Regarding the incident of vehicle C malfunctioning and stopping ahead, according to the rule "if an obstacle is detected ahead and the current state is cruise, determine whether evasive action is necessary," the assessment of the obstacle's impact begins. The assessment is based on factors including the obstacle's location, lane occupancy, distance between the vehicle and the obstacle, and relative speed.

[0093] When evaluating car B's lane-change request, considering that the expected distance after the lane change might fall within the "relatively close" fuzzy set, its membership degree is calculated based on a membership function. Simultaneously, fuzzy factors such as the ease of adjusting its own speed (e.g., the range of acceleration change) are considered, and fuzzy inference is used to determine whether to accept the lane-change request. For example, if the expected distance has a membership degree of 0.6 to the "relatively close" set, and the speed adjustment is within an acceptable range, the lane-change request might be accepted, but the speed needs to be appropriately reduced to maintain safety.

[0094] For a car C facing an obstacle ahead, fuzzy logic is used to assess the urgency of avoidance based on the distance and relative speed to the obstacle. For example, if the distance to the obstacle is 80 meters and the relative speed is 20 km / h, the membership degree of the car at "closer" distance and "rapid approach" speed is calculated separately using the membership function. After comprehensive reasoning, a decision may be made to take a smaller avoidance action, such as moderately veering to the left (within a safe range) and slowing down.

[0095] Based on the combined evaluation results of the two events, the hybrid autopilot of vehicle A decides to switch from cruise control to a combined strategy, which simultaneously handles lane change requests and obstacle avoidance. The new control objectives include: accepting lane changes from vehicle B while maintaining a safe distance from the obstacle vehicle C ahead; constraints cover the safe following distance after lane changes with vehicle B, lateral acceleration limits during obstacle avoidance, and comfort requirements during longitudinal deceleration.

[0096] Through the detailed algorithms described above, the hybrid automaton for fleet cooperation strategies can effectively respond to event information streams. It accurately selects appropriate cooperation strategies by combining rule-based decision trees with fuzzy logic, and achieves smooth strategy switching, ensuring safe, stable, and efficient vehicle operation in complex traffic scenarios. In this process, the event information stream serves as the input to the hybrid automaton, triggering state transitions and strategy switching; the algorithm combining rule-based decision trees and fuzzy logic provides the decision-making basis for the hybrid automaton, realizing a complete process from event perception to strategy execution.

[0097] In this embodiment, "controlling the movement of the vehicle based on the control strategy" specifically includes: When the control strategy is a follow strategy, the following operations are performed: Establish a first dynamic model of the vehicle in the longitudinal direction. In the first dynamic model, the longitudinal acceleration of the vehicle is... The change is The longitudinal position of the vehicle The relationship with speed is ,in, Let be the longitudinal velocity of the car, t be time, m be the mass of the car, and k be a constant; the objective function of the first dynamic model is to minimize the value of J. Where N is a natural number, For the car and the car in front at time The actual distance at that time For the car and the car in front at time Safe distance at time For the car at time The longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at time; the constraint condition of the first dynamic model is: the first preset threshold ≤ acceleration. ≤Second preset threshold, 0≤speed ≤ Third preset threshold, First preset threshold < 0, Second preset threshold > 0; During the first time period, the objective function in the first dynamics model is minimized based on the position and speed of the car and the position and speed of the car in front, thereby obtaining the acceleration of the car; and during the second time period, the motion of the car is controlled based on the acceleration; wherein, both the first time period and the second time period start from the current time, and the length of the first time period is greater than the length of the second time period.

[0098] In practice, k can be a combined coefficient that takes into account air resistance and rolling resistance. The length of the first time period can be 5s, and the length of the second time period can be 0.1s.

[0099] Optionally, the first preset threshold = The second preset threshold = 3 The third preset threshold is 100km / h.

[0100] In this embodiment, the step of "minimizing the objective function in the first dynamics model based on the position and speed of the car and the position and speed of the car in front" specifically includes: Based on the position and speed of the car, and the position and speed of the car in front, the objective function in the first dynamic model is minimized using the gradient descent method or the sequential quadratic programming method.

[0101] Gradient descent is a first-order optimization algorithm. To find a local minimum of a function using gradient descent, it is necessary to iteratively search in the opposite direction of the gradient (or approximate gradient) of the current point on the function, at a predetermined step distance. If the search is iterated in the opposite direction of the gradient, it will approach the local maximum of the function; this process is called gradient ascent.

[0102] Sequential Quadratic Programming (SQP) is an iterative method for solving nonlinear optimization problems. It approximates the solution by decomposing the original nonlinear optimization problem into a series of quadratic programming (QP) subproblems. Each QP subproblem is constructed and solved at the current iteration point to update the iteration point. The SQP algorithm is applied to solving nonlinear optimization problems. By continuously constructing and solving quadratic programming subproblems, the algorithm gradually approaches the optimal solution of the original nonlinear optimization problem. In this embodiment, "controlling the movement of the vehicle based on the control strategy" specifically includes: When the control strategy is a follow strategy, the following operations are performed: A first dynamic model of the vehicle in the longitudinal direction and a second dynamic model in the lateral direction are established. In the second dynamic model, the lateral acceleration of the vehicle is... The change is The position of the car in the lateral direction The relationship with speed is in, Let be the lateral velocity of the car, t be time, and m be the mass of the car; the objective function of the second dynamic model is to minimize the value of J. Where N is a natural number, For the car at time The lateral position at time, y_target is the lateral position of the target lane. For the car at time longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at that time; the constraint condition of the second dynamic model is: the fourth preset threshold ≤ acceleration. ≤ Fifth preset threshold, 0 ≤ speed The threshold value is ≤ the sixth preset threshold, while the fourth, fifth, and sixth preset thresholds are all greater than zero. Here, the summation range of the J value is N time points, and a longitudinal acceleration square term is also added to ensure the smoothness of the longitudinal velocity change. The sixth preset threshold can be... .

[0103] During the first time period, the objective functions in the first and second dynamic models are minimized to obtain the longitudinal acceleration and lateral acceleration of the vehicle; and during the second time period, the motion of the vehicle is controlled based on the acceleration; wherein, both the first and second time periods start from the current time, and the length of the first time period is greater than the length of the second time period.

[0104] Here, the first dynamics model is used to consider the longitudinal motion of the car during lane changing, and the first dynamics model is still used to maintain a safe longitudinal distance from the car in front.

[0105] Here, optimization algorithms such as gradient descent or SQP can be used to minimize the objective function. In the prediction time domain, considering both lateral and longitudinal motion, the corresponding lateral and longitudinal control variable sequences (lateral acceleration and longitudinal acceleration) are calculated. The optimized lateral and longitudinal accelerations are then output as the current control variables to the vehicle's actuators to control the steering and throttle / brake, achieving smooth lane-changing operations.

[0106] Embodiment 2 of the present invention provides a control device for an automobile, comprising the following modules: The data acquisition module is used to acquire the first state data of the vehicle and the second state data of the surrounding vehicles; The strategy generation module is used to determine the control strategy of the vehicle based on the first and second state data. The control strategy includes a following strategy and a lane-changing strategy. A control module is used to control the movement of the vehicle based on the control strategy.

[0107] In this embodiment, the control module is further configured to: When the control strategy is a follow strategy, the following operations are performed: Establish a first dynamic model of the vehicle in the longitudinal direction. In the first dynamic model, the longitudinal acceleration of the vehicle is... The change is The longitudinal position of the vehicle The relationship with speed is ,in, Let be the longitudinal velocity of the car, t be time, m be the mass of the car, and k be a constant; the objective function of the first dynamic model is to minimize the value of J. Where N is a natural number, For the car and the car in front at time The actual distance at that time For the car and the car in front at time Safe distance at time For the car at time The longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at time; the constraint condition of the first dynamic model is: the first preset threshold ≤ acceleration. ≤Second preset threshold, 0≤speed ≤ Third preset threshold, First preset threshold < 0, Second preset threshold > 0; During the first time period, the objective function in the first dynamics model is minimized based on the position and speed of the car and the position and speed of the car in front, thereby obtaining the acceleration of the car; and during the second time period, the motion of the car is controlled based on the acceleration; wherein, both the first time period and the second time period start from the current time, and the length of the first time period is greater than the length of the second time period.

[0108] In this embodiment, the control module is further configured to: based on the position and speed of the car and the position and speed of the car in front, minimize the objective function in the first dynamic model using gradient descent or sequential quadratic programming.

[0109] In this embodiment, the control module is further configured to: When the control strategy is a follow strategy, the following operations are performed: A first dynamic model of the vehicle in the longitudinal direction and a second dynamic model in the lateral direction are established. In the second dynamic model, the lateral acceleration of the vehicle is... The change is The position of the car in the lateral direction The relationship with speed is in, Let be the lateral velocity of the car, t be time, and m be the mass of the car; the objective function of the second dynamic model is to minimize the value of J. Where N is a natural number, For the car at time The lateral position at time, y_target is the lateral position of the target lane. For the car at time longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at that time; the constraint condition of the second dynamic model is: the fourth preset threshold ≤ acceleration. ≤ Fifth preset threshold, 0 ≤ speed ≤The sixth preset threshold, while the fourth, fifth, and sixth preset thresholds are all greater than zero; During the first time period, the objective functions in the first and second dynamic models are minimized to obtain the longitudinal acceleration and lateral acceleration of the vehicle; and during the second time period, the motion of the vehicle is controlled based on the acceleration; wherein, both the first and second time periods start from the current time, and the length of the first time period is greater than the length of the second time period.

[0110] Embodiment 3 of the present invention provides a terminal, including: one or more processors; and a memory storing one or more programs thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the control method according to Embodiment 1.

[0111] Embodiment 4 of the present invention provides a storage medium on which a computer program is stored, and when the program is executed by a processor, it implements the control method according to Embodiment 1.

[0112] It should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0113] The detailed descriptions listed above are merely specific descriptions of feasible embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. All equivalent embodiments or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

Claims

1. A control method for an automobile, characterized in that, Includes the following steps: Acquire the first state data of the vehicle and the second state data of the surrounding vehicles; Based on the first and second state data, the control strategy of the vehicle is determined, and the control strategy includes: a following strategy and a lane-changing strategy; The movement of the vehicle is controlled based on the control strategy.

2. The control method according to claim 1, characterized in that, The phrase "controlling the movement of the vehicle based on the control strategy" specifically includes: When the control strategy is a follow strategy, the following operations are performed: Establish a first dynamic model of the vehicle in the longitudinal direction. In the first dynamic model, the longitudinal acceleration of the vehicle is... The change is The longitudinal position of the vehicle The relationship with speed is ,in, Let be the longitudinal velocity of the car, t be time, m be the mass of the car, and k be a constant; the objective function of the first dynamic model is to minimize the value of J. Where N is a natural number, For the car and the car in front at time The actual distance at that time For the car and the car in front at time Safe distance at time For the car at time The longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at time; the constraint condition of the first dynamic model is: the first preset threshold ≤ acceleration. ≤Second preset threshold, 0≤speed ≤ Third preset threshold, First preset threshold < 0, Second preset threshold > 0; During the first time period, the objective function in the first dynamics model is minimized based on the position and speed of the car and the position and speed of the car in front, thereby obtaining the acceleration of the car; and during the second time period, the motion of the car is controlled based on the acceleration; wherein, both the first time period and the second time period start from the current time, and the length of the first time period is greater than the length of the second time period.

3. The control method according to claim 2, characterized in that, The phrase "minimizing the objective function in the first dynamics model based on the position and speed of the car and the position and speed of the car in front" specifically includes: Based on the position and speed of the car, and the position and speed of the car in front, the objective function in the first dynamic model is minimized using the gradient descent method or the sequential quadratic programming method.

4. The control method according to claim 2, characterized in that, The phrase "controlling the movement of the vehicle based on the control strategy" specifically includes: When the control strategy is a follow strategy, the following operations are performed: A first dynamic model of the vehicle in the longitudinal direction and a second dynamic model in the lateral direction are established. In the second dynamic model, the lateral acceleration of the vehicle is... The change is The position of the car in the lateral direction The relationship with speed is in, Let be the lateral velocity of the car, t be time, and m be the mass of the car; the objective function of the second dynamic model is to minimize the value of J. Where N is a natural number, For the car at time The lateral position at time, y_target is the lateral position of the target lane. For the car at time longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at that time; the constraint condition of the second dynamic model is: the fourth preset threshold ≤ acceleration. ≤ Fifth preset threshold, 0 ≤ speed ≤The sixth preset threshold, while the fourth, fifth, and sixth preset thresholds are all greater than zero; During the first time period, the objective functions in the first and second dynamic models are minimized to obtain the longitudinal acceleration and lateral acceleration of the vehicle; and during the second time period, the motion of the vehicle is controlled based on the acceleration; wherein, both the first and second time periods start from the current time, and the length of the first time period is greater than the length of the second time period.

5. A control device for an automobile, characterized in that, Includes the following modules: The data acquisition module is used to acquire the first state data of the vehicle and the second state data of the surrounding vehicles; The strategy generation module is used to determine the control strategy of the vehicle based on the first and second state data. The control strategy includes a following strategy and a lane-changing strategy. A control module is used to control the movement of the vehicle based on the control strategy.

6. The control device according to claim 5, characterized in that, The control module is also used for: When the control strategy is a follow strategy, the following operations are performed: Establish a first dynamic model of the vehicle in the longitudinal direction. In the first dynamic model, the longitudinal acceleration of the vehicle is... The change is The longitudinal position of the vehicle The relationship with speed is ,in, Let be the longitudinal velocity of the car, t be time, m be the mass of the car, and k be a constant; the objective function of the first dynamic model is to minimize the value of J. Where N is a natural number, For the car and the car in front at time The actual distance at that time For the car and the car in front at time Safe distance at time For the car at time The longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at time; the constraint condition of the first dynamic model is: the first preset threshold ≤ acceleration. ≤Second preset threshold, 0≤speed ≤ Third preset threshold, First preset threshold < 0, Second preset threshold > 0; During the first time period, the objective function in the first dynamics model is minimized based on the position and speed of the car and the position and speed of the car in front, thereby obtaining the acceleration of the car; and during the second time period, the motion of the car is controlled based on the acceleration; wherein, both the first time period and the second time period start from the current time, and the length of the first time period is greater than the length of the second time period.

7. The control device according to claim 6, characterized in that, The control module is also used for: Based on the position and speed of the car, and the position and speed of the car in front, the objective function in the first dynamic model is minimized using the gradient descent method or the sequential quadratic programming method.

8. The control device according to claim 6, characterized in that, The control module is also used for: When the control strategy is a follow strategy, the following operations are performed: A first dynamic model of the vehicle in the longitudinal direction and a second dynamic model in the lateral direction are established. In the second dynamic model, the lateral acceleration of the vehicle is... The change is The position of the car in the lateral direction The relationship with speed is in, Let be the lateral velocity of the car, t be time, and m be the mass of the car; the objective function of the second dynamic model is to minimize the value of J. Where N is a natural number, For the car at time The lateral position at time, y_target is the lateral position of the target lane. For the car at time longitudinal velocity at that time For the car in front at the moment The longitudinal velocity at that time; the constraint condition of the second dynamic model is: the fourth preset threshold ≤ acceleration. ≤ Fifth preset threshold, 0 ≤ speed ≤The sixth preset threshold, while the fourth, fifth, and sixth preset thresholds are all greater than zero; During the first time period, the objective functions in the first and second dynamic models are minimized to obtain the longitudinal acceleration and lateral acceleration of the vehicle; and during the second time period, the motion of the vehicle is controlled based on the acceleration; wherein, both the first and second time periods start from the current time, and the length of the first time period is greater than the length of the second time period.

9. A terminal, characterized in that, include: One or more processors; A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement the control method according to any one of claims 1 to 4.

10. A storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the control method according to any one of claims 1 to 4.

Images