Active lane changing method, system and equipment for port intelligent outer container truck and storage medium

Abstract

The invention discloses an active lane changing method, system and device for a port intelligent outer container truck, and a storage medium, which are applied to application scenes of large load, strong inertia and complex road conditions of the port outer container truck, and realize fusion of a high-definition vector map and a multi-sensor measurement result through real-time camera external parameter calibration and relative map positioning. The distance measurement precision of the long-distance obstacle target is improved; in the planning stage, map constraint and obstacle prediction are fused, aiming at various triggering conditions such as static obstacles, dynamic obstacles, recommended lane changing, forced lane changing points and the like, a lane changing intention is automatically generated or canceled, a safe and efficient planning track is given, and the driving safety of a vehicle is improved; cooperative operation of multiple vehicle networking and a vehicle team is supported, all vehicles in the vehicle team can share real-time positions, speeds and planned tracks through a special communication network to be overall dispatched and optimized, mutual interference among the vehicles is reduced, and a more efficient and safer cluster type driving mode is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent driving, and relates to active lane change in autonomous driving. Specifically, it relates to an active lane change method, system, device and storage medium for port intelligent external container trucks. Background Technique

[0002] In recent years, with the rapid development of autonomous driving technology, intelligent logistics and intelligent transportation have gradually become the core development directions of the global transportation industry. Especially against the background of the rapid development of maritime transportation, the annual average container throughput of ports has exceeded 2000 million TEUs (standard container units), and the efficient operation of port logistics has become increasingly important for the national economy. As an important means of transportation connecting maritime and land transportation and realizing multimodal transportation, port external container trucks (i.e., external container transport trucks at ports) play a key role in the trunk transportation tasks between ports and warehouses. However, traditional manually driven container trucks have problems such as high labor costs, difficult management, and potential safety hazards caused by driver fatigue during operation, and it is difficult to meet the efficient and safe requirements of modern port logistics.

[0003] The introduction of autonomous driving technology provides a new solution for the operation of port external container trucks. By realizing autonomous driving, port external container trucks can significantly improve transportation efficiency, reduce labor and management costs, and reduce traffic accidents caused by human operation errors. However, the application of autonomous driving technology in port external container trucks faces many challenges. Especially in the ever-changing and high-traffic highway environment, how to ensure the safe and stable operation of vehicles has become the focus of continuous attention in the industry.

[0004] During the implementation of autonomous driving, the active lane change function is crucial. Since port external container trucks frequently pass through different road conditions such as highways, suburban roads, and even urban ring roads during operation, the traffic environment has characteristics such as diverse vehicle types, large driving speed ranges, and significant differences in lane numbers and rules. If lane change operations cannot be carried out in a timely and accurate manner, traffic accidents and congestion are likely to occur. In addition, compared with passenger cars, autonomous driving external container trucks have the characteristics of large load and long body. If the lane change operation is carried out too urgently or too late, it may lead to unstable vehicle posture, increased risk of rollover, and even pose a threat to the safety of surrounding vehicles.

[0005] At present, the existing automatic lane-changing technologies for autonomous driving are mainly designed for passenger cars, and their lane-changing strategies and execution methods are difficult to be directly applied to port container trucks. For the few autonomous lane-changing technologies targeting container trucks, they do not fully utilize high-definition vector maps. Specifically, although some technologies use GNSS-RTK for positioning and can project the obstacle detection results in the sensor coordinate system to the high-definition vector map coordinate system, in the actual operation process, it is usually assumed that the external parameters of the sensor remain unchanged after offline calibration, without considering the dynamic changes of the external parameters due to factors such as vehicle vibration and load changes during actual operation, and also ignoring the possible systematic deviation between the high-definition vector map and GNSS-RTK. It should be noted that most of the current existing technologies mainly conduct research based on the assumed scenarios of passenger cars. In actual applications, the special requirements such as "heavy load and strong inertia" presented by port container trucks during operation are not taken into key consideration, and there is a lack of high-precision support for long-distance obstacle detection in high-intensity logistics scenarios. It is difficult to ensure the safety and efficiency of large port container trucks in complex port and road environments. Summary of the Invention

[0006] In view of the above problems, the main object of the present invention is to design an active lane-changing method, system, device and storage medium for intelligent port container trucks, which can solve the lane-changing problem of port container trucks in complex traffic environments and improve the safety and stability of their operation by combining the accurate road information of high-definition vector maps, the environmental perception data of multi-source sensors, and intelligent decision-making and control.

[0007] To achieve the above object, the present invention adopts the following technical solutions: An automatic driving active lane-changing system for port container trucks, the system includes a movable carrier, as well as a multi-source sensor, a microcomputer, and a high-definition vector map arranged on the movable carrier; Movable carrier, the movable carrier is an unmanned port container truck equipped with an intelligent driving system; Multi-source sensor, including a camera, GNSS-RTK, IMU, and chassis wheel speed sensor installed on the movable carrier, the camera includes a 30° narrow-angle camera and a 120° wide-angle camera, which are used to capture the front road image of the movable carrier, and the front road image includes road elements and obstacles; Microcomputer, which is used to run the intelligent driving system. The intelligent driving system is deployed with a perception and detection module, a map positioning module, a prediction and decision-making module, and a planning and control module. Through the collaborative work of multiple modules, it combines the high-definition vector map to predict the movement trajectory of dynamic obstacles and generates the lane-changing intention of the vehicle itself to achieve the active lane-changing of the movable carrier; High-definition vector map, a high-definition map including several map elements, which is used to provide the position information of road elements in the map to the intelligent driving system.

[0008] As a further description of the present invention, the perception detection module is used to detect the front road image through a detection model to obtain the position areas of road elements and obstacles in the front road image; The map positioning module is used to estimate the position, attitude of the movable carrier relative to the high-definition vector map, and the external parameters of the camera according to the results of the perception detection module and the high-definition vector map; The prediction and decision-making module, based on the data of multi-source sensors, the perception detection module, and the results of the map positioning module, combines the lane lines of the high-definition vector map to estimate the position, speed, and attitude of the obstacles in the vehicle body coordinate system, and predicts the movement trajectories of dynamic obstacles; The planning and control module, based on the results of the map positioning module and the prediction and decision-making module, combines the high-definition vector map to generate an active lane-changing intention, determine the lane-changing trajectory, and execute the active lane-changing behavior.

[0009] An active lane-changing method for autonomous driving of port external container trucks, which is based on the above system, specifically includes the following steps: Synchronously collect images of the road in front of the vehicle through a 30° narrow-angle camera and a 120° wide-angle camera, and detect and identify road elements and obstacles on the images to obtain the category and position information of different targets within the detection range on the images. Among them, the road elements are static, including lane lines, road lamp posts, and road signs, and the obstacles are respectively dynamic obstacles and static obstacles, including vehicles and cones; Register and align the static road elements detected on the image with the corresponding static road elements in the high-definition vector map to obtain the position and attitude of the vehicle relative to the high-definition vector map, and correct the external parameters of the camera relative to the vehicle; Based on the position information of the obstacles on the image, the detection results of the lane lines on the image, and the external parameters of the camera, combined with the prior information of the lane lines in the high-definition vector map, perform data fusion through ranging to obtain the ranging information of the obstacles outside the detection range, the position relationship relative to the lane lines, and the position of the obstacles in the vehicle coordinate system, and predict the future movement trajectory based on the historical positions of the dynamic obstacles; Judge whether to generate a lane-changing intention and a lane-changing trajectory through the historical trajectory and predicted trajectory of the obstacles relative to the high-definition vector map, combined with the prior information of the high-definition vector map, and execute the lane-changing behavior.

[0010] As a further description of the present invention, preprocess the images collected by the camera to obtain the category and position information of different targets within 200 meters of the vehicle in the detection range on the images. Specifically, perform target-level detection on the image sequence through a classic network model to output the pixel-level positions of static road elements, dynamic obstacles, and static obstacles.

[0011] As a further description of the present invention, the position and attitude of the vehicle relative to the high-definition vector map are obtained, and the external parameters of the camera relative to the vehicle are corrected, including the following steps: Receive the preprocessed sensor data. Based on the angular velocity and acceleration provided by the IMU and the classical strapdown inertial navigation model, predict the current position and attitude of the vehicle. Combine the image detection results, high-definition map data, and the previous positioning state of the vehicle to correct the current positioning state of the vehicle, and use the continuous positioning states of the vehicle as prior information; According to the prior information of the vehicle positioning state, load the map elements within a preset range around the vehicle in the high-definition vector map and preprocess the map elements; Based on the loaded and preprocessed map elements and the image detection results, perform object-level association of semantic elements through the classical Hungarian method. Calculate the position and attitude of the vehicle in the high-definition map based on the position information of the semantic elements in the high-definition map and the relative position between the vehicle and the semantic elements.

[0012] As a further description of the present invention, the ranging and fusion tracking of obstacles outside the detection range include the following steps: Based on the corrected external parameters of the camera and the prior information of the lane lines in the high-definition map, perform weighted fusion of the distance data through the classical Kalman filtering method to obtain the accurate position of the obstacles outside the detection range; Using the corrected external parameters of the camera, convert the position of the obstacle in the image coordinate system to the vehicle coordinate system. Based on the position and attitude of the vehicle in the high-definition vector map coordinate system, obtain the absolute position of the obstacle in the vehicle coordinate system converted to the high-definition vector map through continuous coordinate transformation; Filter and estimate the state of the historical trajectory of the dynamic obstacle. Adopt the classical vehicle kinematic model to generate the motion trajectory within a preset time window according to the starting speed, position, and orientation of the dynamic obstacle.

[0013] As a further description of the present invention, generating a lane change intention and a lane change trajectory includes the following steps: Obtain the historical trajectory and predicted trajectory information of the obstacle; Based on the predicted trajectory of the dynamic obstacle and the prior information of the high-definition vector map, determine whether a lane change is required. Among them, the situations where a lane change is required are: static obstacle interference, dynamic obstacle interference, recommended lane change, and forced lane change point; Judge the safety of the intended lane change by calculating the time to collision t, and the expression is: t = D / (v0 + v1); Where D is the relative distance between the host vehicle and the obstacle, and v0 and v1 are the speeds of the host vehicle and the obstacle respectively; If the time to collision is greater than the safety threshold, the lane change is considered safe, and a lane change command is sent to the chassis of the host vehicle; If the time to collision is less than or equal to the safety threshold, the lane change is considered unsafe, and no lane change command is sent to the chassis of the host vehicle; The host vehicle planning and control module controls the host vehicle to execute a lane change according to the steering angle and acceleration / deceleration commands provided by the vehicle chassis, and monitors the relative distance between the host vehicle and the vehicle in front in real time during the lane change. If the distance is less than the safety distance threshold, emergency braking or replanning is performed for avoidance.

[0014] An electronic device includes a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete communication with each other through the communication bus. The memory is used to store computer programs; The processor is used to execute the above method by running the computer program stored on the memory.

[0015] A computer-readable storage medium stores a computer program, where the computer program, when executed by a processor, implements the above method.

[0016] Compared with the prior art, the technical effect of the present invention is: The present invention provides an active lane change method, system, device, and storage medium for a port intelligent external container truck, which is applied to the application scenario of large load, strong inertia, and complex road conditions of the port external container truck. Through real-time camera extrinsic calibration and relative map positioning, it realizes the fusion of high-definition vector map and multi-sensor measurement results, improving the ranging accuracy of long-distance obstacle targets; in the planning stage, it fuses map constraints and obstacle prediction, and autonomously generates or cancels lane change intentions and gives safe and efficient planned trajectories for various triggering situations such as static obstacles, dynamic obstacles, recommended lane changes, and forced lane change points, improving the driving safety of the host vehicle; at the same time, it supports multi-vehicle networking and fleet collaborative operation. Each vehicle in the fleet can share real-time positions, speeds, and planned trajectories through a dedicated communication network (such as 5G or V2X) for overall scheduling and optimization, reducing mutual interference between each other, and realizing a more efficient and safer cluster driving mode. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic structural diagram of the system of the present invention; Figure 2 It is a schematic flowchart of the method of the present invention; Figure 3 It is an example diagram of the present invention applied to an actual scenario. DETAILED DESCRIPTION OF THE INVENTION

[0018] The present invention will be described in detail below with reference to the accompanying drawings: In view of the problem of autonomous lane change of port automated guided vehicles (AGVs), this application discloses an active lane change method and system for autonomous driving based on high-definition vector maps and visual target detection. For out-port AGVs with heavy loads and long bodies, it can generate lane change intentions in advance and leave sufficient operation buffer time, enabling out-port AGVs to perform safe and efficient autonomous lane change behaviors.

[0019] In one embodiment of the present invention, an active lane change system for autonomous driving of out-port AGVs is disclosed. Referring to Figure 1 as shown, it includes four parts: a movable carrier, multi-source sensors, a microcomputer, and a high-definition vector map. Among them, the multi-source sensors, the microcomputer, and the high-definition vector map are arranged on the movable carrier, and the multi-source sensor and high-definition vector map data are used as the input of the system. Specifically: The movable carrier is an out-port AGV (hereinafter referred to as "ego vehicle") equipped with an intelligent driving system and is driverless. The multi-source sensors include, but are not limited to, cameras, GNSS-RTK, IMU, and chassis wheel speed sensors installed on the movable carrier. The cameras include a 30° narrow-angle camera and a 120° wide-angle camera, which are arranged at the front end of the vehicle body and are used to capture the front road image of the movable carrier. The front road image includes road elements (such as lane lines, road lamp posts, road signs, etc.) and obstacles (such as vehicles, cones, etc.). The sensors need to be calibrated. The calibrated parameters of the sensors are divided into internal parameters and external parameters of the sensors. Among them, the internal parameters are only related to the physical characteristics of the sensors themselves, such as lens focal length, optical center position, etc.; while the external parameters mainly refer to the relative position relationships between the sensors and between the sensors and the carrier. The microcomputer is used to run the intelligent driving system. The intelligent driving system is deployed with a perception and detection module, a map positioning module, a prediction and decision-making module, and a planning and control module. Through the collaborative work of multiple modules, it predicts the movement trajectories of dynamic obstacles in combination with the high-definition vector map and generates the lane change intention of the ego vehicle to achieve the active lane change of the movable carrier. The high-definition vector map is a high-definition map including several map elements, which is used to provide the position information of road elements in the map to the intelligent driving system. The high-definition map is stored in the hard disk of the microcomputer in a format including, but not limited to, the NDS format. The high-definition vector map represents the road boundary of the road in a vector form, that is, an ordered form of points and lines.

[0020] In this embodiment, each module installed on the microcomputer will be described in detail: The perception and detection module is used to detect the forward road image through a detection model to obtain the position areas of road elements and obstacles in the forward road image; The map positioning module is used to estimate the position, attitude of the movable carrier relative to the high-definition vector map, and the external parameters of the camera according to the results of the perception and detection module and the high-definition vector map; The prediction and decision-making module estimates the position, speed, and attitude of the obstacles in the vehicle body coordinate system based on the data of multi-source sensors, the perception and detection module, and the results of the map positioning module, and combines the lane lines of the high-definition vector map to predict the movement trajectories of dynamic obstacles; The planning and control module generates an active lane-changing intention, determines the lane-changing trajectory, and executes the active lane-changing behavior based on the results of the map positioning module and the prediction and decision-making module, in combination with the high-definition vector map.

[0021] In this embodiment, among the above multi-source sensors, GNSS-RTK provides centimeter-level absolute position information, and combined with IMU, it can achieve high-precision attitude estimation; IMU is used to measure the acceleration and angular velocity of the vehicle to improve the positioning continuity in scenarios with weak GNSS signals; The chassis wheel speed sensor provides real-time vehicle speed information to help the system master the motion state of the vehicle itself.

[0022] Based on the above system, in another embodiment of the present invention, an active lane-changing method for autonomous driving applied to port container trucks is disclosed. Refer to Figure 2 As shown, the method includes the following steps: Step 1: Synchronously collect images of the road in front of the vehicle itself through a 30° narrow-angle camera and a 120° wide-angle camera, and detect and identify road elements and obstacles on the images to obtain the category and position information of different targets within the detection range on the images. Among them, road elements are static, including but not limited to lane lines, road lamp poles, and road signs, and obstacles are respectively dynamic obstacles and static obstacles, including but not limited to vehicles and cones; Step 2: Register and align the static road elements detected on the images with the corresponding static road elements in the high-definition vector map to obtain the position and attitude of the vehicle itself relative to the high-definition vector map, and correct the external parameters of the camera relative to the vehicle itself; Step 3: Based on the position information of the obstacles on the images, the detection results of the lane lines on the images, and the camera external parameters, combined with the prior information of the lane lines in the high-definition vector map, perform data fusion through ranging to obtain the ranging information of the obstacles outside the detection range, the position relationship relative to the lane lines, and the position of the obstacles in the vehicle coordinate system, and predict the future movement trajectories based on the historical positions of the dynamic obstacles; Step 4: Based on the historical and predicted trajectories of the obstacles relative to the high-definition vector map, combined with the prior information of the high-definition vector map, determine whether to generate a lane-changing intention and lane-changing trajectory, and execute the lane-changing behavior.

[0023] Specifically, in this embodiment, the above steps are analyzed in detail as follows: In Step 1, it is implemented through the perception detection module; specifically, a 30° narrow-angle camera and a 120° wide-angle camera are used to synchronously capture the area in front of the vehicle to obtain road scene images, and preprocessing operations such as distortion correction are performed on the images. The internal parameters of the camera required for the undistortion process need to be obtained through the camera calibration process. Common calibration methods include, but are not limited to, Zhang Zhengyou calibration method, Tsai calibration method, etc.

[0024] For the preprocessed images, obtain the category and position information of different targets within 200 meters of the vehicle in the image. Specifically, perform object-level detection on the image sequence through a classic network model, and output the pixel-level positions of static road elements (such as lane lines, lamp posts, road signs, etc.), dynamic obstacles (such as other vehicles, etc.), and static obstacles (such as cones, etc.).

[0025] It should be noted that the obstacle (such as cone, vehicle) detection model selects YOLOv8 as the basic framework. The model output includes the bounding box (box), width, and height of the cone; the lane line detection is based on the SegFormer network architecture, and outputs the semantic segmentation map, class label, and instance segmentation map of the lane line respectively. Subsequent processing includes coordinate transformation, lane line clustering, and noise removal, and at the same time, lane line filtering and tracking are performed. After data fitting, the lane line position information in the image coordinate system and the lane line fitting curve in the vehicle body coordinate system are finally output. The detection process performs inference on multiple ROI regions and can identify targets and lane lines up to 200 meters away.

[0026] In Step 2, it is implemented through the map positioning module by fusing the perception detection module, multi-source sensor data, and high-definition vector map data; specifically, obtain the position and attitude of the vehicle relative to the high-definition vector map, and correct the external parameters of the camera relative to the vehicle, including the following steps: Receive the preprocessed sensor data, and based on the angular velocity and acceleration provided by the IMU, predict the current position and attitude of the vehicle based on the classic strapdown inertial navigation model. Combine the image detection results, high-definition map data, and the previous positioning state of the vehicle to correct the current positioning state of the vehicle, and use the continuous positioning state of the vehicle as prior information (usually, combining GNSS-RTK, IMU, and chassis wheel speed sensor data can provide relatively accurate and continuous prior information on the positioning state of the vehicle).

[0027] Using the pre-calibrated extrinsic parameters of the vehicle sensors, including: the lever arm and installation error angle between the IMU and the vehicle body, the lever arm between the GNSS main antenna and the IMU, the scale factor of the chassis wheel speed, etc.; through the IMU, GNSS-RTK, and chassis vehicle speed, using the method of classical error-state Kalman filtering, fusing data such as the angular velocity, linear acceleration provided by the IMU, the global position provided by GNSS-RTK, and the vehicle speed provided by the chassis vehicle speed, performing inertial integrated navigation, and outputting high-frequency vehicle navigation information. When the subsequent steps fail, the processing results of this step can be sent independently.

[0028] According to the prior information of the vehicle positioning state, load the map elements within a preset range around the vehicle in the high-definition vector map and preprocess the map elements; specifically, through the position and attitude of the vehicle in the map coordinate system currently, project the shape points of the high-definition vector map elements around the vehicle into the vehicle coordinate system. Among them, the x coordinate represents the longitudinal distance of the shape point from the vehicle, the y coordinate represents the lateral distance of the shape point from the vehicle, and the z coordinate represents the height of the shape point in the vehicle coordinate system; according to the x coordinate of the shape points of the road boundary where the vehicle is located in the vehicle coordinate system, only retain the shape points within a certain distance range, such as within the range of 5 meters to 300 meters.

[0029] Based on the loaded and preprocessed map elements and the image detection results, perform object-level association of semantic elements through the classical Hungarian method. Based on the position information of the semantic elements in the high-definition map and the relative position between the vehicle and the semantic elements, infer the position and attitude of the vehicle in the high-definition map. Specifically, according to the road element detection results of the perception detection module, perform semantic-level association on different types of road elements respectively. The association method uses the Hungarian method. Among them, for discrete elements (such as traffic signs), the method of calculating the matching weight is to project the high-definition vector map elements into the image coordinate system through the relative map positioning result and the camera extrinsic parameters, and calculate the Euclidean distance between the center of the vector map and the center of the perception detection result for each element. For lane lines, the perception detection model additionally outputs the detection results in the vehicle coordinate system. Therefore, project the vector map lane lines into the vehicle coordinate system through the relative map positioning result, and sample at a resolution of 1 meter to calculate the average Euclidean distance from the perception detection lane lines as the weight of the Kuhn-Munkres matching algorithm. According to the association relationship of the elements, construct a nonlinear optimization problem. The loss function is the Euclidean distance between the shape points of the map elements in the image coordinate system and the corresponding perception detection results, and use the Gauss-Newton method to solve it to obtain the position and attitude of the vehicle relative to the high-precision map and the correction value of the camera extrinsic parameters.

[0030] In step 3, it is implemented by the prediction and decision-making module based on sensor data, the results of the perception and detection module, the results of the map positioning module, and the high-definition vector map lane lines; specifically, ranging and fusion tracking of obstacles outside the specific detection range (more than 200 meters) includes the following steps: Based on the corrected extrinsic camera parameters and the prior information of the lane lines in the high-definition map, through various ranging methods (such as monocular perspective ranging, stereometric ranging, map-assisted positioning ranging, etc.), according to the classical Kalman filtering method, weighted fusion of distance data is performed to obtain the accurate position of obstacles with a detection range up to 200 meters away; specifically, first calculate the depth information (longitudinal distance) of the obstacle through the obstacle grounding line, use the pixel distance between the obstacle grounding line and the lane line points at the same depth on the image, query the lane line width in the high-definition vector map, and based on the assumption of constant line width and the pixel distance of the obstacle grounding line relative to the lane line, the position of the obstacle in the camera coordinate system can be calculated. Among them, the process of calculating the grounding line can be realized through IPM (Inverse Perspective Mapping).

[0031] Using the corrected extrinsic camera parameters, convert the position of the obstacle in the image coordinate system to the ego-vehicle coordinate system, and based on the position and attitude of the ego-vehicle in the high-definition vector map coordinate system, through continuous coordinate transformation, obtain the absolute position of the obstacle in the ego-vehicle coordinate system converted to the high-definition vector map.

[0032] Filter and estimate the state of the historical trajectory of dynamic obstacles, adopt a classical vehicle kinematic model (such as a bicycle model), and generate a motion trajectory within a preset time window according to the starting speed, position, and orientation of the dynamic obstacle; specifically, filter and estimate the state of the historical trajectory of dynamic obstacles and output its motion trajectory within a preset time window (such as 3s).

[0033] In step 4, it is implemented by the planning and control module based on the results of the map positioning module and the prediction and decision-making module; specifically, generate a lane-changing intention and a lane-changing trajectory, including the following steps: Obtain the historical trajectory and predicted trajectory information of the obstacle; Based on the predicted trajectory of the dynamic obstacle and the prior information of the high-definition vector map, determine whether a lane-changing requirement is generated. Among them, the situations where a lane-changing requirement is generated are: static obstacle interference, dynamic obstacle interference, recommended lane change, and forced lane-changing point; static obstacle interference: such as cones, construction signs occupying the road; dynamic obstacle interference: there are slow, broken-down, or other blocking vehicles on the driving path; recommended lane change: in a multi-lane scenario, the default rightmost lane can be set as the highest-priority lane, and the leftmost lane is prohibited from passing when subject to traffic light rules; forced lane-changing point: for example, when the distance to the ramp is less than a predetermined threshold, the system automatically changes lanes into the specified lane in advance; By calculating the collision time t, the safety of the intended lane change is judged, and the expression is: t = D / (v0 + v1); Where D is the relative distance between the host vehicle and the obstacle, and v0 and v1 are the speeds of the host vehicle and the obstacle respectively; If the collision time is greater than the safety threshold, the lane change is considered safe, and a lane change command is sent to the chassis of the host vehicle; If the collision time is less than or equal to the safety threshold, the lane change is considered unsafe, and no lane change command is sent to the chassis of the host vehicle; The host vehicle planning and control module controls the host vehicle to execute a lane change according to the steering angle and acceleration / deceleration commands provided by the vehicle chassis, and monitors the relative distance between the host vehicle and the vehicle in front in real time during the lane change. If the distance is less than the safety distance threshold, emergency braking or replanning is performed for avoidance. That is, in the case of a change in the environment resulting in an unsafe lane change, emergency braking or replanning is performed.

[0034] As Figure 3 shown, it is an actual case of autonomous lane change operation. The right image is the result taken by a 30° long-distance camera, and it shows that there are traffic cones blocking the road in the distance of the host vehicle's own lane in the image; the left side is the position of the host vehicle and the obstacle relative to the high-definition vector map from an aerial view, as well as the planned guiding line of the host vehicle (the dark thick solid line extending from the front of the vehicle). It can be seen that the guiding line of the host vehicle bypasses the traffic cones and other obstacle vehicles in front.

[0035] Through the above embodiments, the system and method of the present invention are disclosed. Compared with the prior art, the present invention has the following advantages: 1. The active lane change method of the present invention can achieve more efficient and safe driving in application scenarios such as port external container trucks with large load, strong inertia, and complex road conditions; 2. The present invention realizes the fusion of high-definition vector map and multi-sensor measurement results through real-time camera extrinsic parameter calibration and relative map positioning, improving the ranging accuracy of long-distance obstacle targets; 3. The present invention integrates map constraints and obstacle prediction in the planning stage. For various triggering situations such as static obstacles, dynamic obstacles, recommended lane changes, and forced lane change points, it autonomously generates or cancels lane change intentions, and gives safe and efficient planned trajectories, improving the driving safety of the host vehicle; 4. The present invention simultaneously supports multi-vehicle networking and fleet cooperative operation. Each vehicle in the fleet can share real-time positions, speeds, and planned trajectories through a dedicated communication network (such as 5G or V2X) for overall scheduling and optimization, reducing mutual interference between each other, and realizing a more efficient and safer cluster driving mode.

[0036] In another embodiment of the present invention, there is also an electronic device, which may include a processor and a memory storing computer program instructions.

[0037] Specifically, in this embodiment, the above-mentioned processor may include a central processing unit (CPU), or a specific integrated circuit, or may be configured as one or more integrated circuits of this embodiment; the above-mentioned memory may include a mass storage for data or instructions, and for this memory, including but not limited to a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disc, a magneto-optical disc, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these; in appropriate cases, the memory may include removable or non-removable (or fixed) media; in a specific embodiment, the memory is a non-volatile solid-state memory. In a specific embodiment, the memory includes a read-only memory (ROM). In appropriate cases, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or a flash memory, or a combination of two or more of these.

[0038] The above-mentioned processor realizes the active lane-changing method disclosed above in the present invention by reading and executing the computer program instructions stored in the memory.

[0039] It should also be noted that the electronic device of this embodiment may further include a communication interface and a communication bus. Among them, the processor, the memory, and the communication interface are connected through the communication bus and complete communication with each other. The communication interface is mainly used to realize the communication between each unit, each module, device or equipment in the embodiments of the present invention.

[0040] The above-mentioned communication bus includes hardware, software, or a combination of both, and couples the components of the online data traffic device to each other. In appropriate cases, the communication bus may include one or more buses.

[0041] In addition, in combination with the active lane-changing method in the above-mentioned embodiments, an embodiment of the present invention can be implemented by providing a computer storage medium, on which computer program instructions are stored; the computer program instructions are executed by the processor to perform the above-mentioned active lane-changing method.

[0042] It should be clear that the present invention is not limited to the above-disclosed methods, systems, and devices, and also includes various changes, modifications, and additions made by those skilled in the art based on the ideas of the present invention, or changes in the order between steps.

[0043] When implemented in hardware, the present invention may be an electronic circuit, an application specific integrated circuit, appropriate firmware, a plug-in, a functional card, etc.; when implemented in software, the elements of the present invention are programs or code segments used to perform the required tasks. The programs or code segments may be stored in a machine-readable medium, or uploaded via a data signal carried in a carrier wave over a transmission medium or a communication link. The "machine-readable medium" may include any medium capable of storing or transmitting information, such as: electronic circuits, semiconductor memory devices, ROMs, flash memories, erasable ROMs (EROMs), floppy disks, optical discs, hard disks, optical fiber media, radio frequency links, etc. The code segments may be downloaded via a computer network such as the Internet, an intranet, etc.

[0044] The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Any other modifications or equivalent replacements made by those of ordinary skill in the art to the technical solutions of the present invention shall be covered within the scope of the claims of the present invention as long as they do not depart from the spirit and scope of the technical solutions of the present invention.

Claims

1. An autonomous driving active lane-changing system applied to port external container trucks, characterized in that The system includes a movable carrier, as well as a multi-source sensor, a microcomputer, and a high-definition vector map deployed on the movable carrier; The movable carrier is an unmanned port container truck outside the port equipped with an intelligent driving system; The multi-source sensor includes a camera, GNSS-RTK, IMU, and chassis wheel speed sensor installed on the movable carrier. The camera includes a 30° narrow-angle camera and a 120° wide-angle camera, which are used to capture the road image in front of the movable carrier. The road image in front includes road elements and obstacles; The microcomputer is used to run the intelligent driving system. The intelligent driving system is deployed with a perception and detection module, a map positioning module, a prediction and decision-making module, and a planning and control module. Through the collaborative work of multiple modules, combined with the high-definition vector map, it predicts the movement trajectory of dynamic obstacles and generates a lane-changing intention for the vehicle itself to achieve the active lane change of the movable carrier; The high-definition vector map is a high-definition map including several map elements, which is used to provide the position information of road elements in the map to the intelligent driving system.

2. The automatic driving active lane-changing system for port external container trucks according to claim 1, wherein: The perception and detection module is used to detect the road image in front through a detection model to obtain the position areas of road elements and obstacles in the road image in front; The map positioning module is used to estimate the position, attitude of the movable carrier relative to the high-definition vector map, and the external parameters of the camera according to the results of the perception and detection module and the high-definition vector map; The prediction and decision-making module, based on the data of the multi-source sensor, the perception and detection module, and the results of the map positioning module, combined with the lane lines of the high-definition vector map, estimates the position, speed, and attitude of the obstacle in the vehicle body coordinate system, and predicts the movement trajectory of the dynamic obstacle; The planning and control module, based on the results of the map positioning module and the prediction and decision-making module, combined with the high-definition vector map, generates an active lane-changing intention, determines the lane-changing trajectory, and executes the active lane-changing behavior.

3. An automatic driving active lane-changing method for port external container trucks applied to the system according to any one of claims 1-2, characterized in that, The method includes the following steps: Synchronously collect images of the road in front of the vehicle through a 30° narrow-angle camera and a 120° wide-angle camera, and detect and identify road elements and obstacles on the images to obtain the category and position information of different targets within the detection range on the images. Among them, the road elements are static, including lane lines, road lamp poles, and road signs. The obstacles are respectively dynamic obstacles and static obstacles, including vehicles and cones; Register and align the static road elements detected on the image with the corresponding static road elements in the high-definition vector map to obtain the position and attitude of the vehicle relative to the high-definition vector map, and correct the external parameters of the camera relative to the vehicle; Based on the position information of the obstacle on the image, the detection results of the lane lines on the image, and the external parameters of the camera, combined with the prior information of the lane lines of the high-definition vector map, perform data fusion through ranging to obtain the ranging information of the obstacle outside the detection range, the position relationship relative to the lane line, and the position of the obstacle in the vehicle coordinate system, and predict the future movement trajectory based on the historical position of the dynamic obstacle; Based on the historical trajectory and predicted trajectory of the obstacle relative to the high-definition vector map, combined with the prior information of the high-definition vector map, determine whether to generate a lane-changing intention and a lane-changing trajectory, and execute the lane-changing behavior.

4. An active lane-changing method for autonomous driving applied to port external container trucks according to claim 3, characterized in that: Preprocess the images collected by the camera to obtain the category and position information of different targets within 200 meters of the vehicle itself in the images. Specifically, perform object-level detection on the image sequence through a classic network model, and output the pixel-level positions of static road elements, dynamic obstacles, and static obstacles.

5. The automatic driving active lane-changing method for external container trucks at ports according to claim 3, characterized in that: Obtain the position and attitude of the vehicle itself relative to the high-definition vector map, and correct the external parameters of the camera relative to the vehicle itself, including the following steps: Receive the preprocessed sensor data. Based on the angular velocity and acceleration provided by the IMU and the classic strapdown inertial navigation model, predict the current position and attitude of the vehicle itself. Combine the image detection results, high-definition map data, and the previous positioning state of the vehicle itself to correct the current positioning state of the vehicle itself, and use the continuous positioning state of the vehicle itself as prior information. According to the prior information of the vehicle's positioning state, load the map elements within a preset range around the vehicle itself in the high-definition vector map, and preprocess the map elements. Based on the loaded and preprocessed map elements and the image detection results, perform object-level association of semantic elements through the classic Hungarian method. Calculate the position and attitude of the vehicle itself in the high-definition map based on the position information of the semantic elements in the high-definition map and the relative position between the vehicle itself and the semantic elements.

6. The active lane-changing method for autonomous driving applied to port external container trucks according to claim 3, wherein: Range measurement and fusion tracking of obstacles outside the detection range, including the following steps: Based on the corrected external parameters of the camera and the prior information of the lane lines in the high-definition map, perform weighted fusion of the distance data through the classic Kalman filter method to obtain the accurate position of the obstacles outside the detection range. Use the corrected external parameters of the camera to convert the position of the obstacle in the image coordinate system to the vehicle coordinate system. Based on the position and attitude of the vehicle itself in the high-definition vector map coordinate system, obtain the absolute position of the obstacle in the high-definition vector map after continuous coordinate transformation. Filter and estimate the state of the historical trajectory of dynamic obstacles. Adopt the classic vehicle kinematic model to generate a motion trajectory within a preset time window according to the starting speed, position, and orientation of the dynamic obstacle.

7. The automatic driving active lane-changing method for out-port container trucks according to claim 3, wherein: Generate a lane-changing intention and a lane-changing trajectory, including the following steps: Obtain the historical trajectory and predicted trajectory information of the obstacle. Based on the predicted trajectory of the dynamic obstacle and the prior information of the high-definition vector map, determine whether a lane-changing requirement is generated. Among them, the situations where a lane-changing requirement is generated are: static obstacle interference, dynamic obstacle interference, recommended lane change, and forced lane-changing point. Judge the safety of the lane to be changed by calculating the time to collision t. The expression is: t = D / (v0 + v1); Where D is the relative distance between the vehicle itself and the obstacle, and v0 and v1 are the speeds of the vehicle itself and the obstacle respectively. If the time to collision is greater than the safety threshold, it is considered that the lane change is safe, and a lane-changing instruction is sent to the vehicle chassis. If the time to collision is less than or equal to the safety threshold, it is considered that the lane change is unsafe, and no lane-changing instruction is sent to the vehicle chassis. The self-vehicle planning and control module controls the self-vehicle to execute a lane change according to the steering angle and acceleration / deceleration commands provided by the vehicle chassis, and monitors the relative distance between the self-vehicle and the vehicle in front in real time during the lane change. If the distance is less than the safety distance threshold, emergency braking or replanning is performed for avoidance.

8. An electronic device, comprising a processor, a communication interface, a memory, and a communication bus, wherein, The processor, the communication interface, and the memory complete their mutual communication through the communication bus. It is characterized in that the memory is used to store computer programs. The processor is used to execute the method described in any one of claims 3-7 by running the computer program stored on the memory.

9. A computer-readable storage medium, characterized in that, A computer program is stored in the storage medium, wherein the computer program, when executed by the processor, implements the method described in any one of claims 3-7.

Images