Automatic driving path optimization control method and device integrating environment perception and prediction

Abstract

The invention discloses an automatic driving path optimization control method and device fusing environmental perception and prediction. The method comprises the following steps: acquiring environmental perception data of an automatic driving vehicle through a radar and a camera; the initial posture, the target posture and the environment perception data of the automatic driving vehicle serve as road constraint conditions, and a preliminary driving path of the automatic driving vehicle from the current position to the target position is planned through a path planning algorithm; state estimation and prediction are carried out on surrounding vehicles and obstacles through a multi-target tracking algorithm; according to future motion trails of surrounding vehicles and obstacles, road constraint conditions and physical restrictions of the autonomous vehicle, performing path optimization on the initial path to obtain an optimized driving path; and converting the optimized driving path into a control instruction of the automatic driving vehicle, and driving the automatic driving vehicle to stably drive along the optimized driving path through the control instruction. According to the invention, the safety and the driving efficiency of the automatic driving vehicle are further improved.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to an autonomous driving path optimization control method and device, and a computing device that integrates environmental perception and prediction. Background Art

[0002] With the rapid development of autonomous driving technology, the requirements for environmental perception, path planning and control capabilities of autonomous driving vehicles are also increasing.

[0003] Existing path planning algorithms often only consider the static environment and ignore the dynamic changes of surrounding vehicles and obstacles, resulting in a less than optimal planned path and even safety hazards. The driving path cannot be adjusted in time according to environmental changes. When converting the path planning results into control instructions, there are often lags and inaccuracies, which prevents the vehicle from driving smoothly along the planned path.

[0004] To solve the above problems, the present invention proposes an autonomous driving path optimization control method that integrates environmental perception and prediction to further improve the safety and driving efficiency of autonomous driving vehicles. Summary of the invention

[0005] In view of the above problems, the present invention provides an automatic driving path optimization control method and device, and a computing device that integrate environmental perception and prediction.

[0006] According to one aspect of the present invention, there is provided an automatic driving path optimization control method integrating environment perception and prediction, comprising: Acquire environmental perception data of the autonomous driving vehicle through radar and camera, wherein the environmental perception data includes the position, speed, acceleration, lane markings, and obstacle positions of the autonomous driving vehicle and its surrounding vehicles; Using the starting posture, target posture and environmental perception data of the autonomous driving vehicle as road constraints, and planning a preliminary driving path of the autonomous driving vehicle from a current position to a target position through a path planning algorithm; The state of the surrounding vehicles and obstacles is estimated and predicted by a multi-target tracking algorithm to obtain the future motion trajectories of the surrounding vehicles and obstacles; According to the future motion trajectories of the surrounding vehicles and obstacles, the road constraints and the physical limitations of the autonomous driving vehicle, the preliminary path is optimized to obtain an optimized driving path; wherein the physical limitations include a maximum speed and a minimum turning radius; the objectives of the path optimization include minimizing the driving time, maximizing the path smoothness and the minimum safe distance from the surrounding vehicles and obstacles; The optimized driving path is converted into control instructions for the autonomous driving vehicle, and the autonomous driving vehicle is driven to drive smoothly along the optimized driving path by the control instructions, wherein the control instructions include steering angle, acceleration and braking force.

[0007] In an optional manner, the radar includes a forward radar, a rearward radar, a side radar, and a top 360-degree radar; The cameras include a front view camera, a rear view camera, a side view camera, a panoramic camera and an infrared camera.

[0008] In an optional manner, the path planning algorithm incorporates hybrid Algorithms and dynamic window algorithms; Among them, the hybrid The algorithm is used to plan a preliminary path from the starting pose to the goal pose on the global map; The dynamic window algorithm is used to select the optimal control input within a preset time window based on the vehicle's current speed, acceleration and deceleration.

[0009] In an optional manner, the multi-target tracking algorithm uses a detection and tracking learning network; In the state estimation stage, the Kalman filter is used to smooth the motion state of each target and the Hungarian algorithm is used for data association; In the state prediction stage, the state vector and covariance matrix of the next moment are calculated according to the CAM constant acceleration model; The calculation formula of the CAM constant acceleration model is: in, and are the state vector and covariance matrix of the target at time t, respectively. The state vector is , is the position coordinate, is the velocity component, is the acceleration component; is the state transfer matrix from time t to time t+1; B is the control input matrix; is the control input vector; is the process noise covariance matrix.

[0010] In an optional manner, the objective function of the path optimization is: Where N is the length of the prediction time domain; is the travel time of the kth time step; is a measure of path smoothness; To maintain a safe distance from surrounding vehicles and obstacles; , , They are the adjustment coefficients for travel time, path smoothness and safety distance respectively.

[0011] In an optional manner, the calculation formula for the steering angle is: in, is the steering angle; is the vehicle wheelbase; is the curvature of the path; is the slope of the tangent line at the point where the vehicle is located on the path; The calculation formulas of the acceleration are: in, is the acceleration; is the expected speed; is the current speed; is the acceleration distance; The calculation formulas for the braking force are: in, For braking force; is the friction coefficient; is the positive pressure of the vehicle on the ground or the weight of the vehicle; The maximum speed limit for the vehicle.

[0012] In an optional manner, the method further includes: According to the environmental perception data, the motion status of surrounding vehicles and obstacles, and traffic rules, the Markov decision method is used to dynamically adjust the priority of control instructions.

[0013] In an optional manner, the detection and tracking learning network includes a detection module, a prediction module, an association module and a tracking cycle management module; The detection module uses a YOLOv5 network to extract the target position, shape and category of the target information; The prediction module uses a CAM constant acceleration model to estimate the state at the next moment based on the vehicle's current position, speed, acceleration, and covariance matrix; The association module is used to use the Hungarian algorithm to associate and match the target detected in the current frame with the target tracked in the previous frame to obtain multiple matching bipartite graphs, wherein the weight of the bipartite graph is the intersection-union ratio between the detection frame and the prediction frame; The tracking cycle management module is used to regard the target extracted by the detection module as a target to be tracked with a life cycle, wherein when the target just appears, an identification and a state model are created for it; when the number of frames in which the target disappears from the screen exceeds a preset range, the identification and the state model are destroyed.

[0014] According to another aspect of the present invention, there is provided an automatic driving path optimization control device integrating environment perception and prediction, comprising: An environment perception acquisition module, used to acquire environment perception data of the autonomous driving vehicle through radar and camera sensors, wherein the environment perception data includes the position, speed, acceleration, lane markings, and obstacle positions of the autonomous driving vehicle and its surrounding vehicles; A preliminary path planning module, used to plan a preliminary driving path of the autonomous driving vehicle from a current position to a target position by using a path planning algorithm, taking the starting posture, target posture and the environmental perception data of the autonomous driving vehicle as road constraints; A multi-target tracking and prediction module is used to estimate and predict the states of the surrounding vehicles and obstacles through a multi-target tracking algorithm to obtain the future motion trajectories of the surrounding vehicles and obstacles; a path optimization module, configured to optimize the preliminary path to obtain an optimized driving path according to the future motion trajectories of the surrounding vehicles and obstacles, the road constraints, and the physical limitations of the autonomous driving vehicle; wherein the physical limitations include a maximum speed and a minimum turning radius; and the objectives of path optimization include minimizing driving time, maximizing path smoothness, and minimizing safe distances from surrounding vehicles and obstacles; A control instruction generation and execution module is used to convert the optimized driving path into control instructions for the autonomous driving vehicle, and drive the autonomous driving vehicle to drive smoothly along the optimized driving path through the control instructions, wherein the control instructions include steering angle, acceleration and braking force.

[0015] According to another aspect of the present invention, there is provided a computing device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus.

[0016] According to the solution provided by the present invention, environmental perception data of the autonomous driving vehicle is obtained through radar and camera, wherein the environmental perception data includes the position, speed, acceleration, lane markings and obstacle position of the autonomous driving vehicle and its surrounding vehicles; the starting posture, target posture and the environmental perception data of the autonomous driving vehicle are used as road constraints, and a preliminary driving path of the autonomous driving vehicle from the current position to the target position is planned through a path planning algorithm; the state of the surrounding vehicles and obstacles is estimated and predicted through a multi-target tracking algorithm to obtain the future motion trajectory of the surrounding vehicles and obstacles; according to the future motion trajectory of the surrounding vehicles and obstacles, the road constraints and the physical limitations of the autonomous driving vehicle, the preliminary path is optimized to obtain an optimized driving path; wherein the physical limitations include maximum speed and minimum turning radius; the goals of path optimization include minimizing driving time, maximizing path smoothness and minimum safe distance from surrounding vehicles and obstacles; the optimized driving path is converted into a control instruction of the autonomous driving vehicle, and the autonomous driving vehicle is driven to drive smoothly along the optimized driving path through the control instruction, wherein the control instruction includes steering angle, acceleration and braking force. The present invention uses 360-degree forward, backward, side and top radars, as well as forward, rear, side, panoramic and infrared cameras to obtain environmental perception data around the autonomous driving vehicle in all directions and at multiple angles. The algorithm and dynamic window algorithm are used for path planning, which not only ensures the global optimality of the path, but also takes into account the real-time and dynamic characteristics. The detection and tracking learning network is used for multi-target tracking, which further accurately and reliably tracks the future motion trajectories of surrounding vehicles and obstacles. The priority of control instructions is dynamically adjusted through the Markov decision method, which further improves the stability of the autonomous driving system. The tracking cycle management module intelligently manages the life cycle of the target to be tracked, which improves the efficiency and accuracy of multi-target tracking.

[0017] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings: Figure 1A schematic diagram showing a flow chart of an automatic driving path optimization control method integrating environment perception and prediction according to an embodiment of the present invention; Figure 2 A schematic diagram of a radar coverage area according to an embodiment of the present invention is shown; Figure 3 A schematic diagram of path optimization according to an embodiment of the present invention is shown; Figure 4 A schematic diagram of a parking lot scene according to an embodiment of the present invention is shown; Figure 5A A schematic diagram of path planning for a parking lot scenario according to an embodiment of the present invention is shown; Figure 5B A schematic diagram of path optimization in a parking lot scenario according to an embodiment of the present invention is shown; Fig. 6A A schematic diagram showing the direction of a vehicle on a path before optimization according to an embodiment of the present invention is shown; Figure 6B A schematic diagram showing an optimized direction of a vehicle on a path according to an embodiment of the present invention; Fig. 7A A panoramic diagram showing dynamic detection and planning of an embodiment of the present invention Figure 1 ; Figure 7B A panoramic diagram showing dynamic detection and planning of an embodiment of the present invention Figure 2 ; Figure 7C A panoramic diagram showing dynamic detection and planning of an embodiment of the present invention Figure 3 ; Figure 8 A schematic diagram of a framework of an automatic driving path optimization control device integrating environment perception and prediction according to an embodiment of the present invention is shown; Fig. 9 A schematic diagram of the structure of a computing device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0019] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0020] Figure 1 The flowchart of the automatic driving path optimization control method integrating environment perception and prediction according to an embodiment of the present invention is shown. Figure 1 As shown, the following steps are included: Step S101, obtaining environmental perception data of the autonomous driving vehicle through radar and camera, wherein the environmental perception data includes the position, speed, acceleration, lane markings and obstacle positions of the autonomous driving vehicle and its surrounding vehicles.

[0021] In this embodiment, the radar includes a forward radar, a rear radar, a side radar, and a top 360-degree radar. The camera includes a front camera, a rear camera, a side camera, a panoramic camera, and an infrared camera. Figure 2 As shown, the lane marking lines, radar coverage area, and clustered detection objects are obtained with the help of radar and camera.

[0022] Step S102, using the starting posture, target posture and environmental perception data of the autonomous driving vehicle as road constraints, and planning a preliminary driving path of the autonomous driving vehicle from the current position to the target position through a path planning algorithm.

[0023] In this embodiment, the path planning algorithm integrates hybrid Algorithms and dynamic window algorithms, such as Figure 3 , Figure 4 , Figure 5A As shown, hybrid The algorithm is used to plan a preliminary path from the starting posture to the target posture on the global map; the dynamic window algorithm is used to select the optimal control input within a preset time window based on the vehicle's current speed, acceleration, and deceleration. The algorithm and dynamic window algorithm can not only plan the global optimal path, but also make real-time adjustments in the local range. The planned path is more in line with the actual driving capability of the vehicle.

[0024] Step S103, using a multi-target tracking algorithm to estimate and predict the states of the surrounding vehicles and obstacles, and obtain future motion trajectories of the surrounding vehicles and obstacles.

[0025] In this embodiment, the multi-target tracking algorithm adopts a detection and tracking learning network. In the state estimation stage, the Kalman filter is used to smooth the motion state of each target, and the Hungarian algorithm is combined for data association. In the state prediction stage, the state vector and covariance matrix of the next moment are calculated according to the CAM constant acceleration model. The calculation formula of the CAM constant acceleration model is: in, and are the state vector and covariance matrix of the target at time t, respectively. The state vector is , is the position coordinate, is the velocity component, is the acceleration component; is the state transfer matrix from time t to time t+1; B is the control input matrix; is the control input vector; is the process noise covariance matrix.

[0026] In this embodiment, the detection and tracking learning network includes a detection module, a prediction module, an association module, and a tracking cycle management module; Among them, the detection module uses the YOLOv5 network to extract the target position, shape and category of the target information; The prediction module uses the CAM constant acceleration model to estimate the state at the next moment based on the vehicle's current position, velocity, acceleration, and covariance matrix; The association module is used to use the Hungarian algorithm to associate and match the target detected in the current frame with the target tracked in the previous frame to obtain multiple matching bipartite graphs, wherein the weight of the bipartite graph is the intersection-union ratio between the detection frame and the prediction frame; The tracking cycle management module is used to regard the target extracted by the detection module as a target to be tracked with a life cycle, wherein when the target just appears, an identification and a state model are created for it; when the number of frames in which the target disappears from the screen exceeds a preset range, the identification and the state model are destroyed.

[0027] Specifically, if FIG. 7A to FIG. 7C As shown in the figure, the detection module uses the YOLOv5 network to perform real-time target detection on the input image or video frame, extract the position (bounding box coordinates), shape (aspect ratio) and category (vehicle, pedestrian, obstacle, etc.) of the target. The prediction module uses the CAM constant acceleration model to predict the state according to the current state (position, velocity, acceleration) and covariance matrix of the target output by the detection module, and calculates the target state at the next moment. The association module associates and matches the target detected in the current frame with the target tracked in the previous frame, calculates the intersection over union (IoU) between the detection frame and the prediction frame using the Hungarian algorithm, and selects the best match. The tracking cycle management module assigns a unique identifier and state model to each detected target, and records its state changes during its life cycle. When the target disappears from the screen for more than a preset number of frames (such as 5 frames or 10 frames), it is considered that the target has left the field of view, and its identifier and state model are destroyed to release resources and avoid unnecessary calculations.

[0028] Step S104, optimizing the preliminary path to obtain an optimized driving path according to the future motion trajectories of the surrounding vehicles and obstacles, the road constraints and the physical limitations of the autonomous driving vehicle; wherein the physical limitations include a maximum speed and a minimum turning radius; and the objectives of path optimization include minimizing driving time, maximizing path smoothness and minimizing a safe distance from surrounding vehicles and obstacles.

[0029] like Figure 7C As shown in the figure, the autonomous vehicle is driving on the highway. There is a curve ahead and a large truck is driving slowly in the left lane. At this time, the path optimization algorithm first uses sensors to sense and identify the current position and speed information of the large truck and predict its future movement trajectory. At the same time, according to the road constraints (such as speed limit, curve curvature, etc.) and the physical limitations of the autonomous vehicle (such as maximum speed, minimum turning radius, etc.), Figure 5B , Fig. 6A , Figure 6B As shown in the figure, the initial path is optimized to maintain a safe distance and avoid collisions with large trucks, while maximizing the path smoothness and reducing the number of sharp turns, ensuring the stability and safety of the autonomous vehicle on curves, while reducing driving time and energy consumption. During driving, if the speed of the large truck changes or other obstacles appear in front, the path will be dynamically adjusted and optimized in a timely manner, and the autonomous vehicle can flexibly respond to various complex scenarios.

[0030] In an optional manner, the objective function of the path optimization is: Where N is the length of the prediction time domain; is the travel time of the kth time step; is a measure of path smoothness; To maintain a safe distance from surrounding vehicles and obstacles; , , They are the adjustment coefficients for travel time, path smoothness and safety distance respectively.

[0031] In this embodiment, the path optimization problem is converted into an optimization problem for solving the minimization of the objective function. The objective function can be solved by the gradient descent method, Newton's method, etc. to obtain the optimized driving path, which is not limited in this article.

[0032] Step S105: Convert the optimized driving path into control instructions for the autonomous driving vehicle, and drive the autonomous driving vehicle to drive smoothly along the optimized driving path through the control instructions, wherein the control instructions include steering angle, acceleration, and braking force.

[0033] In this embodiment, the optimized driving path is converted into control instructions to ensure that the autonomous driving vehicle drives smoothly along the optimized path. The generation and execution of control instructions are highly real-time and can generate control instructions that meet actual needs according to different road environments, traffic conditions and vehicle performance parameters to quickly respond to road changes and emergencies and ensure driving safety.

[0034] Specifically, the optimized driving path is parsed to extract path parameters such as curvature and slope. According to the path parameters and vehicle performance parameters, the path is converted into a series of control instructions (steering angle, acceleration, braking force). The PID control algorithm is used to ensure the smoothness and continuity of the control instructions to avoid drastic acceleration, deceleration or steering of the vehicle. The control instructions are sent to the vehicle control system to drive the vehicle along the optimized driving path.

[0035] In an optional manner, the calculation formula for the steering angle is: in, is the steering angle; is the vehicle wheelbase; is the curvature of the path; is the slope of the tangent line at the point where the vehicle is located on the path; The calculation formulas of the acceleration are: in, is the acceleration; is the expected speed; is the current speed; is the acceleration distance; The calculation formulas for the braking force are: in, For braking force; is the friction coefficient; is the positive pressure of the vehicle on the ground or the weight of the vehicle; The maximum speed limit for the vehicle.

[0036] In this embodiment, the required steering angle of the vehicle is accurately calculated based on the curvature and tangent slope of the path, reducing unnecessary steering operations. The reasonable acceleration of the vehicle is determined based on the expected speed and the current speed, so that the vehicle can accelerate smoothly and avoid the discomfort caused by sudden acceleration. The required braking force is calculated based on the current speed and the maximum speed limit to ensure that the vehicle can stop safely in an emergency. For example, in a complex urban road environment, the vehicle needs to turn and accelerate and decelerate frequently. The above formula can be used to dynamically adjust according to the real-time road conditions to ensure the safety and stability of the vehicle. In addition, by controlling the braking force and steering angle, the wear of the tires and steering mechanism can be reduced, and the replacement frequency and maintenance cost can be reduced.

[0037] In an optional manner, the method further includes: According to the environmental perception data, the motion status of surrounding vehicles and obstacles, and traffic rules, the Markov decision method is used to dynamically adjust the priority of control instructions.

[0038] In this embodiment, dynamic programming is performed through the Markov decision method to maximize the cumulative reward in the long term, thereby finding the optimal strategy. In other words, the vehicle can make the current optimal decision based on the future state and avoid short-sighted behavior. Integrating traffic rules into the reward function and state transition probability will help improve the social acceptance and safety of autonomous vehicles. For example, the vehicle's sensors sense that there are pedestrians crossing the road in front, and calculate the priority of different control instructions (such as deceleration, steering, etc.) based on the perceived environmental data, state transition probability and reward function. In the above example, deceleration has the highest priority because deceleration can directly avoid collisions with pedestrians.

[0039] According to the solution provided by the present invention, environmental perception data of the autonomous driving vehicle is obtained through radar and camera, wherein the environmental perception data includes the position, speed, acceleration, lane markings and obstacle position of the autonomous driving vehicle and its surrounding vehicles; the starting posture, target posture and the environmental perception data of the autonomous driving vehicle are used as road constraints, and a preliminary driving path of the autonomous driving vehicle from the current position to the target position is planned through a path planning algorithm; the state of the surrounding vehicles and obstacles is estimated and predicted through a multi-target tracking algorithm to obtain the future motion trajectory of the surrounding vehicles and obstacles; according to the future motion trajectory of the surrounding vehicles and obstacles, the road constraints and the physical limitations of the autonomous driving vehicle, the preliminary path is optimized to obtain an optimized driving path; wherein the physical limitations include maximum speed and minimum turning radius; the goals of path optimization include minimizing driving time, maximizing path smoothness and minimum safe distance from surrounding vehicles and obstacles; the optimized driving path is converted into a control instruction of the autonomous driving vehicle, and the autonomous driving vehicle is driven to drive smoothly along the optimized driving path through the control instruction, wherein the control instruction includes steering angle, acceleration and braking force. The present invention uses 360-degree forward, backward, side and top radars, as well as forward, rear, side, panoramic and infrared cameras to obtain environmental perception data around the autonomous driving vehicle in all directions and at multiple angles. The algorithm and dynamic window algorithm are used for path planning, which not only ensures the global optimality of the path, but also takes into account the real-time and dynamic characteristics. The detection and tracking learning network is used for multi-target tracking, which further accurately and reliably tracks the future motion trajectories of surrounding vehicles and obstacles. The priority of control instructions is dynamically adjusted through the Markov decision method, which further improves the stability of the autonomous driving system. The tracking cycle management module intelligently manages the life cycle of the target to be tracked, which improves the efficiency and accuracy of multi-target tracking.

[0040] Figure 8 The schematic diagram of the framework of the automatic driving path optimization control device integrating environmental perception and prediction according to an embodiment of the present invention is shown. The automatic driving path optimization control device integrating environmental perception and prediction comprises: An environment perception acquisition module 810 is used to acquire environment perception data of the autonomous driving vehicle through radar and camera sensors, wherein the environment perception data includes the position, speed, acceleration, lane markings, and obstacle positions of the autonomous driving vehicle and its surrounding vehicles; A preliminary path planning module 820 is used to plan a preliminary driving path of the autonomous driving vehicle from a current position to a target position by using a path planning algorithm, taking the starting position, the target position and the environmental perception data of the autonomous driving vehicle as road constraints; A multi-target tracking and prediction module 830 is used to estimate and predict the states of the surrounding vehicles and obstacles through a multi-target tracking algorithm to obtain the future motion trajectories of the surrounding vehicles and obstacles; A path optimization module 840 is used to optimize the preliminary path to obtain an optimized driving path according to the future motion trajectories of the surrounding vehicles and obstacles, the road constraints and the physical limitations of the autonomous driving vehicle; wherein the physical limitations include a maximum speed and a minimum turning radius; and the objectives of path optimization include minimizing driving time, maximizing path smoothness and a minimum safe distance from surrounding vehicles and obstacles; The control instruction generation and execution module 850 is used to convert the optimized driving path into the control instructions of the autonomous driving vehicle, and drive the autonomous driving vehicle to drive smoothly along the optimized driving path through the control instructions, wherein the control instructions include steering angle, acceleration and braking force.

[0041] Fig. 9 The schematic diagram of the structure of the computing device embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the computing device.

[0042] like Fig. 9 As shown, the computing device may include: a processor (processor) 902 , a communication interface (Communications Interface) 904 , a memory (memory) 906 , and a communication bus 908 .

[0043] The processor 902, the communication interface 904, and the memory 906 communicate with each other via the communication bus 908. The communication interface 904 is used to communicate with other devices such as a client or other server network elements. The processor 902 is used to execute the program 910, which can specifically execute the relevant steps in the above-mentioned embodiment of the automatic driving path optimization control method integrating environmental perception and prediction.

[0044] Specifically, the program 910 may include program codes, which include computer operation instructions.

[0045] The processor 902 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the computing device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0046] The memory 906 is used to store the program 910. The memory 906 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0047] According to the solution provided by the present invention, environmental perception data of the autonomous driving vehicle is obtained through radar and camera, wherein the environmental perception data includes the position, speed, acceleration, lane markings and obstacle position of the autonomous driving vehicle and its surrounding vehicles; the starting posture, target posture and the environmental perception data of the autonomous driving vehicle are used as road constraints, and a preliminary driving path of the autonomous driving vehicle from the current position to the target position is planned through a path planning algorithm; the state of the surrounding vehicles and obstacles is estimated and predicted through a multi-target tracking algorithm to obtain the future motion trajectory of the surrounding vehicles and obstacles; according to the future motion trajectory of the surrounding vehicles and obstacles, the road constraints and the physical limitations of the autonomous driving vehicle, the preliminary path is optimized to obtain an optimized driving path; wherein the physical limitations include maximum speed and minimum turning radius; the goals of path optimization include minimizing driving time, maximizing path smoothness and minimum safe distance from surrounding vehicles and obstacles; the optimized driving path is converted into a control instruction of the autonomous driving vehicle, and the autonomous driving vehicle is driven to drive smoothly along the optimized driving path through the control instruction, wherein the control instruction includes steering angle, acceleration and braking force. The present invention uses 360-degree forward, backward, side and top radars, as well as forward, rear, side, panoramic and infrared cameras to obtain environmental perception data around the autonomous driving vehicle in all directions and at multiple angles. The algorithm and dynamic window algorithm are used for path planning, which not only ensures the global optimality of the path, but also takes into account the real-time and dynamic characteristics. The detection and tracking learning network is used for multi-target tracking, which further accurately and reliably tracks the future motion trajectories of surrounding vehicles and obstacles. The priority of control instructions is dynamically adjusted through the Markov decision method, which further improves the stability of the autonomous driving system. The tracking cycle management module intelligently manages the life cycle of the target to be tracked, which improves the efficiency and accuracy of multi-target tracking.

[0048] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into multiple submodules or subunits or subcomponents. All features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature that provides the same, equivalent or similar purpose. In addition, those skilled in the art will appreciate that, although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and to form different embodiments. For example, in the following claims, any one of the claimed embodiments may be used in any combination. The present invention may be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim that lists several means, several of these means may be embodied by the same hardware item. The steps in the above embodiments should not be understood as limiting the execution order unless otherwise specified.

Claims

1. An automatic driving path optimization control method integrating environment perception and prediction, characterized in that: include: Acquire environmental perception data of the autonomous driving vehicle through radar and camera, wherein the environmental perception data includes the position, speed, acceleration, lane markings, and obstacle positions of the autonomous driving vehicle and its surrounding vehicles; Using the starting posture, target posture and environmental perception data of the autonomous driving vehicle as road constraints, and planning a preliminary driving path of the autonomous driving vehicle from a current position to a target position through a path planning algorithm; The state of the surrounding vehicles and obstacles is estimated and predicted by a multi-target tracking algorithm to obtain the future motion trajectories of the surrounding vehicles and obstacles; According to the future motion trajectories of the surrounding vehicles and obstacles, the road constraints and the physical limitations of the autonomous driving vehicle, the preliminary path is optimized to obtain an optimized driving path; wherein the physical limitations include a maximum speed and a minimum turning radius; the objectives of the path optimization include minimizing the driving time, maximizing the path smoothness and the minimum safe distance from the surrounding vehicles and obstacles; The optimized driving path is converted into control instructions for the autonomous driving vehicle, and the autonomous driving vehicle is driven to drive smoothly along the optimized driving path by the control instructions, wherein the control instructions include steering angle, acceleration and braking force.

2. The automatic driving path optimization control method integrating environment perception and prediction according to claim 1 is characterized in that: The radar includes a forward radar, a rear radar, a side radar and a top 360-degree radar; The cameras include a front view camera, a rear view camera, a side view camera, a panoramic camera and an infrared camera.

3. The automatic driving path optimization control method integrating environment perception and prediction according to claim 1 is characterized in that: The path planning algorithm integrates hybrid Algorithms and dynamic window algorithms; Among them, the hybrid The algorithm is used to plan a preliminary path from the starting pose to the goal pose on the global map; The dynamic window algorithm is used to select the optimal control input within a preset time window based on the vehicle's current speed, acceleration and deceleration.

4. The automatic driving path optimization control method integrating environment perception and prediction according to claim 1 is characterized in that: The multi-target tracking algorithm adopts a detection and tracking learning network; In the state estimation stage, the Kalman filter is used to smooth the motion state of each target and the Hungarian algorithm is used for data association; In the state prediction stage, the state vector and covariance matrix of the next moment are calculated according to the CAM constant acceleration model; The calculation formula of the CAM constant acceleration model is: in, and are the state vector and covariance matrix of the target at time t, respectively. The state vector is , is the position coordinate, is the velocity component, is the acceleration component; is the state transfer matrix from time t to time t+1; B is the control input matrix; is the control input vector; is the process noise covariance matrix.

5. The automatic driving path optimization control method integrating environment perception and prediction according to claim 1 is characterized in that: The objective function of the path optimization is: Where N is the length of the prediction time domain; is the travel time of the kth time step; is a measure of path smoothness; To maintain a safe distance from surrounding vehicles and obstacles; , , They are the adjustment coefficients for travel time, path smoothness and safety distance respectively.

6. The automatic driving path optimization control method integrating environment perception and prediction according to claim 1 is characterized in that: The calculation formula of the steering angle is: in, is the steering angle; is the vehicle wheelbase; is the curvature of the path; is the slope of the tangent line at the point where the vehicle is located on the path; The calculation formulas of the acceleration are: in, is the acceleration; is the expected speed; is the current speed; is the acceleration distance; The calculation formulas for the braking force are: in, For braking force; is the friction coefficient; is the positive pressure of the vehicle on the ground or the weight of the vehicle; The maximum speed limit for the vehicle.

7. The automatic driving path optimization control method integrating environment perception and prediction according to claim 1 is characterized in that: The method further comprises: According to the environmental perception data, the motion status of surrounding vehicles and obstacles, and traffic rules, the Markov decision method is used to dynamically adjust the priority of control instructions.

8. The automatic driving path optimization control method integrating environment perception and prediction according to claim 4 is characterized in that: The detection and tracking learning network includes a detection module, a prediction module, an association module and a tracking cycle management module; The detection module uses a YOLOv5 network to extract the target position, shape and category of the target information; The prediction module adopts the CAM constant acceleration model to estimate the state at the next moment according to the current position, speed and acceleration of the vehicle and the covariance matrix; The association module is used to use the Hungarian algorithm to associate and match the target detected in the current frame with the target tracked in the previous frame to obtain multiple matching bipartite graphs, wherein the weight of the bipartite graph is the intersection-union ratio between the detection frame and the prediction frame; The tracking cycle management module is used to regard the target extracted by the detection module as a target to be tracked with a life cycle, wherein when the target just appears, an identification and a state model are created for it; when the number of frames in which the target disappears from the screen exceeds a preset range, the identification and the state model are destroyed.

9. An automatic driving path optimization control device integrating environment perception and prediction, characterized in that: include: An environment perception acquisition module, used to acquire environment perception data of the autonomous driving vehicle through radar and camera sensors, wherein the environment perception data includes the position, speed, acceleration, lane markings, and obstacle positions of the autonomous driving vehicle and its surrounding vehicles; A preliminary path planning module, used to plan a preliminary driving path of the autonomous driving vehicle from a current position to a target position by using a path planning algorithm, taking the starting posture, target posture and the environmental perception data of the autonomous driving vehicle as road constraints; A multi-target tracking and prediction module is used to estimate and predict the states of the surrounding vehicles and obstacles through a multi-target tracking algorithm to obtain the future motion trajectories of the surrounding vehicles and obstacles; a path optimization module, configured to optimize the preliminary path to obtain an optimized driving path according to the future motion trajectories of the surrounding vehicles and obstacles, the road constraints, and the physical limitations of the autonomous driving vehicle; wherein the physical limitations include a maximum speed and a minimum turning radius; and the objectives of path optimization include minimizing driving time, maximizing path smoothness, and minimizing safe distances from surrounding vehicles and obstacles; A control instruction generation and execution module is used to convert the optimized driving path into control instructions for the autonomous driving vehicle, and drive the autonomous driving vehicle to drive smoothly along the optimized driving path through the control instructions, wherein the control instructions include steering angle, acceleration and braking force.

10. A computing device comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the above-mentioned autonomous driving path optimization control method that integrates environmental perception and prediction.

Images