An Automatic U-turn Control Method, Device and Driverless Vehicle

Abstract

The present invention discloses an automatic turning control method, device and unmanned vehicle for an unmanned vehicle. The method includes: during the process of the unmanned vehicle driving along the required task path, when it is recognized that a turn is needed, controlling the vehicle to track the task path and go straight until it reaches the turning section, and then performing a path following turn; during the path following turn, when an obstacle is recognized, searching for an obstacle-free path in the same direction as the turning end point according to the task path as the target path, and obtaining a set of target points from the searched target path; iteratively searching for a feasible path based on the obtained target points until the turning is completed; the device includes a processor and a memory, the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method; the unmanned vehicle includes an unmanned vehicle body, and the above device is further provided on the unmanned vehicle body. The present invention has the advantages of simple implementation method, high turning success rate, strong flexibility and environmental adaptability, etc.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned vehicle control, and in particular, to an automatic turning control method, device and unmanned vehicle for an unmanned vehicle. Background Art

[0002] During the automatic driving process of an unmanned vehicle, the driving path is determined through path planning. Path planning is to find a collision-free path from the starting position to the target position based on a certain optimization criterion. Compared with a straight path, it is more complex for an unmanned vehicle to complete a U-turn. In the prior art, the U-turn of an unmanned vehicle is usually achieved by using a tracing method, that is, making the movement trajectory of the unmanned vehicle as close as possible to the tracing points of the pre-set U-turn section, so as to achieve the purpose of U-turn. For example, Chinese Patent Application CN113104053 discloses an automatic turning tracing method and an unmanned vehicle. This solution is to generate two tracing point sets with opposite directions based on the tracing points at the same position when generating the tracing point set. The unmanned vehicle uses the first-direction tracing point set among them as the initial tracing line for tracing. When receiving a turning instruction during the tracing process, if there is no obstacle point within the preset range, it starts to turn until the fourth tracing point in the second tracing point set is found within the preset range. Using this tracing point as the target point, it traces according to the second-direction tracing point set to achieve automatic turning and reverse tracing, enabling the unmanned vehicle to adapt to more driving environments. This solution is based on the tracing method to make the movement trajectory of the unmanned vehicle as close as possible to the pre-set tracing points to complete the U-turn.

[0003] However, the above method for completing the automatic U-turn of an unmanned vehicle based on the tracing method has the following problems:

[0004] 1. It must rely on the tracing path and is only applicable to scenarios where there are no obstacles on the tracing path. It cannot adapt to complex environments. When there are obstacles at the tracing points, the unmanned vehicle cannot accurately track the target path and the U-turn fails.

[0005] 2. Due to the high dependence on the tracing path, it has high requirements for the accuracy of the unmanned vehicle chassis control. If the actual walking trajectory of the unmanned vehicle deviates from the expected path, the U-turn is likely to fail. Summary of the Invention

[0006] The technical problem to be solved by the present invention is: aiming at the technical problems existing in the prior art, the present invention provides an automatic turning control method, device and unmanned vehicle for an unmanned vehicle with a simple implementation method, high U-turn success rate, strong flexibility and environmental adaptability.

[0007] To solve the above technical problems, the technical solution proposed by the present invention is:

[0008] An automatic turning control method for an unmanned vehicle, comprising:

[0009] During the process of the driverless vehicle driving along the required task path, when it is recognized that a U-turn is needed, the vehicle is controlled to track the task path and go straight until it reaches the U-turn section, and then perform a path following turn.

[0010] During the path following turn, when an obstacle is recognized, a clear path in the same direction as the U-turn end point is searched according to the task path as the target path, and a set of target points is obtained from the searched target path.

[0011] A feasible path is iteratively searched based on the obtained target points until the U-turn is completed.

[0012] Further, when it is recognized that a U-turn is needed, the task path is also divided into a first straight section, a U-turn section, and a second straight section. The first straight section is the straight line segment between the driving starting point where it is recognized that a U-turn is needed and the starting point of the U-turn section. The U-turn section is the turning section at the U-turn intersection. The second straight section is the straight line segment between the end point of the U-turn section and the position point where the driverless vehicle completes the U-turn.

[0013] Further, when it is recognized that a U-turn is needed, the driverless vehicle is controlled to go straight along the first straight section until the starting point of the U-turn section.

[0014] When it is recognized that the driverless vehicle has reached the starting point of the U-turn section, the driverless vehicle is controlled to perform a path following turn along the U-turn section with the minimum turning radius.

[0015] When an obstacle is recognized during the path following turn, the driverless vehicle is controlled to enter the path search state to search the target path according to the second straight section.

[0016] Further, the search for the target path according to the second straight section includes: using the second straight section as a reference line, generating a set of parallel reference lines by translating left and right in the lane, and selecting an obstacle-free path closest to the second straight section from the parallel reference lines as the finally selected target path.

[0017] Further, the iterative search for a feasible path based on the obtained target points includes: using the search method of the hybrid A* algorithm to iteratively search for a feasible path to reach each target point from far to near. During the search using the hybrid A* algorithm, the cost of the current node includes the cumulative distance value from the starting point to the current node and the cumulative obstacle distance from the starting point to the current node.

[0018] Further, the cost of the current node is obtained by weighting the cumulative distance value and the cumulative obstacle distance.

[0019] Further, after iteratively searching for a feasible path based on each of the obtained target points, the method further includes: segmenting the searched feasible path according to the movement direction of the path, and controlling the driverless vehicle to enter a path tracking state. During the segmented path tracking process, a smooth path that meets the preset constraint requirements is generated with the segmented path as the target path, and the smooth path generated for each segmented path is tracked until the direction of the driverless vehicle is the same as the direction of the U-turn end point to complete the U-turn or the end point of the last path segment is tracked.

[0020] Further, during the segmented path tracking process, when an obstacle is recognized, the driverless vehicle is controlled to switch to a path search state to search for a feasible path from the current position to the end point.

[0021] A driverless vehicle automatic U-turn control device includes a processor and a memory. The memory is used to store a computer program, and the processor is used to execute the computer program. The processor is used to execute the computer program to perform the method as described above.

[0022] A driverless vehicle includes a driverless vehicle body, and the above-mentioned driverless vehicle automatic U-turn control device is further provided on the driverless vehicle body.

[0023] Compared with the prior art, the advantages of the present invention are as follows:

[0024] 1. During the U-turn process of the driverless vehicle, the present invention first tracks the task path and goes straight until the U-turn section, and then performs path tracking and turning. When an obstacle is recognized, an obstacle-free path with the same direction as the U-turn end point is searched according to the task path as the target path, and a set of target points is obtained on the target path for iterative search of the feasible path. As a result, a path that can avoid obstacles and is closer to the driving habits of humans can be obtained, thereby effectively improving the environmental adaptability, movement flexibility, and U-turn success rate of the driverless vehicle.

[0025] 2. The present invention further iteratively searches for the feasible path for the determined target points by adopting the search method of the hybrid A* algorithm. At the same time, on the basis of the traditional hybrid A* algorithm, a new expected obstacle distance cost is introduced into the cost of the current node, which can make the cost of the node closer to the obstacle higher, so that the planning result tends to be far from the obstacle, and solves the problem that the traditional hybrid A* algorithm will have a problem of being too close to the obstacle when directly searching for the path.

[0026] 3. The present invention further uses the second straight section as a reference line, generates a set of parallel reference lines by translating left and right in the lane, and selects an obstacle-free path closest to the first straight section from each parallel reference line as the finally selected target path, and selects a set of target points on the target path, which can select target points that are as far away from the obstacle as possible and keep the same direction as the task path, and can improve the U-turn success rate.

[0027] 4. Further, in the present invention, for the feasible paths searched out, the paths are segmented according to the movement direction, and each sub-path is used to generate a smooth curve from the current position to the target path in a curve smoothing manner, which can reduce the accuracy requirements for chassis control, and has a certain tolerance space. At the same time, it has a certain memory ability for obstacles in the sensor blind area, and can improve the collision caused by the sensor blind area. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 is a schematic diagram of the implementation process of the automatic U-turn control method for the unmanned vehicle in this embodiment.

[0029] Figure 2 is a schematic diagram of the principle of dividing the U-turn task path into stages in this embodiment.

[0030] Figure 3 is a schematic diagram of the state machine control process for realizing the automatic U-turn control of the unmanned vehicle in the specific application embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] The present invention will be further described below in conjunction with the accompanying drawings of the specification and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.

[0032] As Figure 1 shown, the steps of the automatic U-turn control method for the unmanned vehicle in this embodiment include:

[0033] S01. During the process of the unmanned vehicle driving along the required task path, when it is recognized that a U-turn is needed, control the vehicle to track the task path straight until reaching the U-turn section, and then perform a path following turn.

[0034] S02. During the path following turn, when an obstacle is recognized, search for an obstacle-free path with the same direction as the U-turn end point as the target path according to the task path, and obtain a set of target points on the searched target path.

[0035] S03. Iteratively search for a feasible path according to the obtained target points until the U-turn is completed.

[0036] When a person drives a vehicle to make a U-turn, the reversing direction is usually to make the front of the vehicle face the same direction as the end point, which can reduce the number of reverses. Referring to the driving habit of a person driving a vehicle, in this embodiment, when the unmanned vehicle is in a U-turn intersection or needs to make a U-turn in place due to a task, etc., it first tracks the task path straight until reaching the U-turn section and then performs a path following turn. When an obstacle is recognized, simulate the driving habit of a person driving a vehicle and search for an obstacle-free path with the same direction as the U-turn end point as the target path according to the task path, and obtain a set of target points on the target path for iterative search of a feasible path, so as to obtain a path result that can not only avoid obstacles but also be closer to the driving habit of a person driving a vehicle, thereby effectively improving the environmental adaptability, movement flexibility and U-turn success rate of the unmanned vehicle.

[0037] In this embodiment, sensors are pre-arranged on the driverless vehicle to collect traffic information in the real environment during the driving of the driverless vehicle, and a high-precision map is constructed according to the collected traffic information. The above traffic information includes lanes, curbs, intersections, etc.; the obtained traffic information is transmitted to the local planning module of the driverless vehicle for path planning analysis, which can help make up for the perception blind area and also help improve the local planning module's understanding of the environment. After obtaining the task and the map, a rough task path is generated, and then steps S01-S03 are executed to perform a U-turn task according to the task and the environment. The U-turn task may be that when driving to a U-turn intersection and a U-turn is required, the road conditions ahead will be detected in real time during the driving of the driverless vehicle. When it is detected that the vehicle is at a specified distance from the U-turn intersection, it is determined that a U-turn task needs to be executed, and steps S01-S03 are switched to be executed to complete the U-turn; the U-turn task may also be to make a U-turn in place according to the task requirements. After receiving the in-place U-turn instruction, a U-turn path is first planned, and then steps S01-S03 are switched to be executed to complete the U-turn.

[0038] In this embodiment, when it is recognized that a U-turn is needed, the task path is further divided into a first straight section, a U-turn section, and a second straight section. The first straight section is a straight line segment between the driving starting point where it is recognized that a U-turn is needed and the starting point of the U-turn section. The U-turn section is a turning section at the U-turn intersection. The second straight section is a straight line segment between the end point of the U-turn section and the position point where the driverless vehicle completes the U-turn. Subsequently, the U-turn task is processed in stages according to the task and the environment to simulate the driving habits of a human driver to obtain the driving path, making the U-turn control more reasonable.

[0039] In this embodiment, in step S01, when it is recognized that a U-turn is needed, the driverless vehicle is controlled to go straight along the first straight section to the starting point of the U-turn section; when it is recognized that the driverless vehicle has reached the starting point of the U-turn section, the driverless vehicle is controlled to perform a tracking turn along the U-turn section with the minimum turning radius. As Figure 2 shown, the task path that needs to make a U-turn is first divided into 3 sections: section A (the first straight section), section B (the U-turn section), and section C (the second straight section). When the driverless vehicle recognizes that there is a U-turn intersection at a specified distance ahead, before the driverless vehicle is about to enter the U-turn intersection, it first tracks and goes straight along section A, restricting the obstacle avoidance range not to exceed section B. When the vehicle reaches near the junction of section A and section B, the route is gradually changed from section A to section B in an equally spaced interpolation manner; during the process from section A to section B, a precise line-following method is adopted. When an obstacle (such as a physical obstacle affecting the driving of the driverless vehicle or a virtual wall marked on the map) is recognized during the tracking turn, it jumps from the line-following state to the path search state to search for the target path according to the second straight section until the driverless vehicle turns to the posture of completing the U-turn.

[0040] Preferably, during the straight - line driving of the above - mentioned tracking section A, referring to the driving habits of the driver, the left - and - right detour range is restricted to make the vehicle drive straight along the roadside as much as possible until the starting point of the U - turn section at the U - turn intersection. When following the line for too long in section B, it is preferable to use the arc of the minimum turning - radius path as the forward path in a way of following the track.

[0041] Since the driverless vehicle drives straight along the first straight - line section before entering the U - turn section and the end - point direction after the U - turn is the direction of the second straight - line section, choosing a barrier - free path that is as close as possible to the direction of the second straight - line section and at the same time the closest to the second straight - line section can ensure the effective completion of the U - turn. In this embodiment, searching for the target path according to the second straight - line section specifically includes: taking the second straight - line section as the reference line, generating a set of parallel reference lines by translating left and right in the lane, and selecting an obstacle - free path that is the closest to the second straight - line section from each parallel reference line as the finally selected target path. Then, a set of target points is selected from the searched target path at a certain distance interval, and the value of the interval distance can be specifically configured according to actual needs.

[0042] Take, for example, Figure 2 As shown, when the driverless vehicle recognizes an obstacle during the turning and following - the - track process in section B, section C is used as the reference line, and a set of parallel reference lines parallel to section C is generated by translating left and right in the lane. The paths that meet the following conditions are screened out from each parallel reference line: 1. The closest to section C; 2. Without obstacles and no collision will occur. The path screened out is used as the target path. The target points selected based on this target path can ensure that the selected target points are as far away from the obstacle as possible and keep the same direction as the task path. The selection of the target points is the key to the success of the U - turn, which can effectively improve the success rate of the U - turn.

[0043] The above screening of the obstacle - free path from a set of parallel reference lines can be achieved by detecting obstacles for each parallel path.

[0044] It can be understood that the above - mentioned way of segmenting the task path of the U - turn task and the control method for each segment can also be adaptively implemented in other ways. The key lies in referring to the driving habits of the driver, first following the track and turning according to the task path, and then searching for an obstacle - free path with the same direction as the U - turn end - point direction according to the position of the obstacle and the position of the end - point direction when encountering an obstacle.

[0045] In step S03 of this embodiment, iteratively searching for a feasible path based on the obtained target points specifically includes: using the search method of the hybrid A* algorithm to iteratively search for the feasible paths to reach each target point from far to near. During the search using the hybrid A* algorithm, the cost of the current node includes the cumulative distance value from the starting point to the current node and the cumulative obstacle distance from the starting point to the current node. The selection of target points has relatively high requirements for the planning algorithm based on search. In this embodiment, a target point preprocessing method is used to find a target point that is far from obstacles and has a direction consistent with the subsequent task after a U-turn.

[0046] The hybrid A* algorithm is a point-to-point planning algorithm. Similar to the A* algorithm, it is planned based on a grid map. Different from the A* algorithm, the hybrid A* algorithm considers the kinematic constraints of the vehicle when exploring surrounding nodes and can limit the direction during node expansion. In the hybrid A* algorithm, the cost of a node consists of the path cost of the current node and the heuristic value, that is, f(x) = g(x) + h(x), where x represents the current node, g(x) is the cost of the current node, which is obtained by accumulating the distance values each time an expansion occurs, that is, the cumulative distance from the starting point to the current node, and h(x) is the heuristic value from the current node to the end point, usually the distance value from the current node to the end point. That is, the traditional hybrid A* algorithm directly uses the path length as the cost g(x) of the node, which will result in too close a distance to the obstacle when performing path search.

[0047] In this embodiment, by using the hybrid A* algorithm to iteratively search for the feasible paths for a set of target points determined in step S02, at each node, two arcs with the minimum turning radius and three splines for straight-line forward movement are used, and the same is true for backward movement. At the same time, based on the traditional hybrid A* algorithm, a new cumulative obstacle distance is introduced as the expected obstacle distance cost in the cost of the current node, which can make the cost of nodes close to the obstacle higher, so that the planning result tends to be far from the obstacle, and solves the problem that the traditional hybrid A* algorithm will have too close a distance to the obstacle when directly performing path search.

[0048] In this embodiment, the cost g(x) of the current node is obtained by weighting the cumulative distance value and the cumulative obstacle distance. The specific calculation formula for the cost g(x) of the current node is:

[0049]

[0050] Among them, is the cumulative distance from the starting point x0 to x k , is the cumulative obstacle distance from the starting point x0 to x k , only the values less than the expected obstacle are accumulated, and ω0 and ω1 are the weights corresponding to the cumulative distance and the cumulative obstacle distance respectively.

[0051] When using the hybrid A* algorithm to iteratively search for a feasible path for a set of target points in this embodiment, the path cost of the current node is calculated according to the above formula (1), and the cost of the node expansion direction is determined based on the end orientation, which can make the planned path far away from obstacles. At the same time, by adjusting the cost weights of the path length and the expected obstacle distance, that is, adjusting the coefficients ω0 and ω1 in the above formula (1), the path search effect can also be adjusted, so as to obtain a more reasonable path result.

[0052] Since the hybrid A* algorithm is a point-to-point planning algorithm, the selection of target points will directly affect the success rate and efficiency of U-turn. In this embodiment, by combining the surrounding obstacle information and map information, with the second straight section (section C) as the reference line, a set of parallel reference lines are generated by translating left and right in the lane. Collision detection is started from the second straight section (section C) to select a collision-free path closest to the second straight section (section C), and a set of target points are obtained by sampling the path at equal intervals. A set of target points that are as far away from obstacles as possible and consistent with the direction of the task path are selected. When expanding the reverse nodes near the starting point, only the reverse nodes with the same direction as the end point are considered. In this way, a path result closer to the driving habit of humans can be obtained. Then, combined with the search method of the hybrid A*, it is possible to keep away from obstacles as much as possible during the U-turn process, complete the U-turn efficiently, and improve the U-turn success rate at the same time.

[0053] After iteratively searching for a feasible path based on the obtained target points in this embodiment, it further includes: segmenting the searched feasible path according to the movement direction of the path, and controlling the unmanned vehicle to enter the segmented tracking state. During the segmented tracking process, a smooth path that meets the preset constraint requirements (such as vehicle kinematic constraints) is generated with the segmented path as the target path, and the smooth path generated for each segmented path is tracked until the direction of the unmanned vehicle is the same as the U-turn end direction to complete the U-turn or the end point of the last segment of the path is tracked. When an obstacle is recognized during the segmented tracking process, the unmanned vehicle is controlled to switch to the path search state to search for a feasible path from the current position to the end point.

[0054] After a feasible path is searched by using the search method of the hybrid A* in this embodiment, the path is segmented according to the movement direction. Each sub-path uses the curve smoothing method to generate a smooth curve from the current position to the target path, which can reduce the accuracy requirements for chassis control, and has a certain tolerance space. At the same time, it has a certain memory ability for obstacles in the sensor blind area, and improves the collision caused by the sensor blind area.

[0055] It can be understood that other search methods other than the hybrid A* algorithm can also be used to iteratively search for a feasible path for the target points.

[0056] In a specific application embodiment, the switching of the above-mentioned states can be achieved by designing a state machine, enabling the driverless vehicle to automatically switch between path tracking and random search within the area, and trying to follow the task path as much as possible. The state machine control is as follows Figure 3 As shown, after starting the task execution, it is judged whether there is a U-turn intersection (or a U-turn-in-place task). If so, it is controlled to enter the path tracking state, and the vehicle tracks along the side and goes straight to the starting point of the U-turn section B, that is, the end point of the previous section A before entering the current U-turn section; then it enters the path tracking and turning state, and makes a path tracking turn with the minimum turning radius; during the path tracking turn, it is judged whether there is an obstacle that cannot move forward. If so, it is controlled to enter the path search state, search for an obstacle-free path as the target path with section C as the reference line, and obtain a set of target points on the target path; use the hybrid A* search method to iteratively search for a feasible path for the target points. When a feasible path is found, it is controlled to enter the segmented path tracking state, segment the path, and generate a smooth path that meets the vehicle kinematic constraints with the current segmented path as the target. If it has not reached the end point of the segmented path, continue to execute the next segment until the U-turn is successful; when an obstacle is recognized during the segmented path tracking, it is controlled to switch the driverless vehicle to the path search state to search for a feasible path from the current position to the end point.

[0057] The following takes the realization of the automatic U-turn of the driverless vehicle in a specific application embodiment as an example to further illustrate the present invention.

[0058] The detailed steps for realizing the automatic U-turn control of the driverless vehicle in this embodiment are as follows:

[0059] First, a high-precision map of the running environment of the driverless vehicle is created by the high-precision mapping module, and environmental information such as lanes, intersections, and lane reference lines is depicted on the map; the high-precision positioning module obtains the pose information of the vehicle body in the map from sensors such as GPS and IMU; the perception fusion module obtains the surrounding environment and obstacle information of the vehicle body from sensors such as lidar and cameras; the global task module gives the navigation path that the driverless vehicle needs to travel.

[0060] When the driverless vehicle parses the U-turn section in the global task path, the task path is segmented as shown in Figure 2 into: straight section A, U-turn section B, and straight section C. Before the driverless vehicle is about to enter the U-turn intersection, the left and right detour ranges are restricted, and it goes straight along the roadside as much as possible to the starting point of the U-turn intersection.

[0061] In the U-turn section B, use the path arc with the minimum turning radius as the forward path in a path tracking manner until it encounters an obstacle that cannot move or meets the posture for completing the U-turn. If it stops due to an obstacle, the state switches from path tracking to the path free search state.

[0062] In the free path search state, using the high-precision map and task information, a set of parallel paths of the task path are constructed within the lane range. Obstacle detection is performed on each parallel path, and the obstacle-free path closest to the task route (section C) is selected. A set of target points are selected at a certain distance interval, and the hybrid A* search method is used to iteratively search for the path to reach this set of target points from far to near. When using the hybrid A* search method, the cost of the current node considers the obstacle information and the traffic information given by the high-precision map, and is calculated according to the above formula (1). A feasible path is searched within the passable range.

[0063] After a feasible path is searched, the path is segmented according to the movement direction, and the tracking state is entered again. Using the sub-path as the target path, a smooth path that satisfies the vehicle kinematic model is generated from the current vehicle body position, and each segment of the path is tracked until the direction of the unmanned vehicle is the same as the direction of the U-turn end point to complete the U-turn or the end point of the last segment of the path is tracked, then the task state is entered. During the segmented tracking process, if an obstacle is encountered and cannot be bypassed, the state is converted to free search, and a feasible path to a suitable end point is continued to be searched.

[0064] Embodiment 2:

[0065] The unmanned vehicle automatic U-turn control device in this embodiment includes a processor and a memory. The memory is used to store a computer program, and the processor is used to execute the computer program. The processor is used to execute the computer program to execute the unmanned vehicle automatic U-turn control method. In the process of the unmanned vehicle driving according to the required task path, when it is recognized that a U-turn is needed, it controls to track the task path straight until the U-turn section and then perform a tracking turn; during the tracking turn, when an obstacle is recognized, an obstacle-free path with the same direction as the U-turn end point is searched according to the task path as the target path, and a set of target points are obtained from the searched target path; a feasible path is iteratively searched according to the obtained target points until the U-turn is completed, so as to obtain a path result that can not only avoid obstacles but also be closer to the driving habits of humans, thereby effectively improving the environmental adaptability, movement flexibility and U-turn success rate of the unmanned vehicle. The above unmanned vehicle automatic U-turn control method is specifically referred to as shown in Embodiment 1 and will not be elaborated here one by one.

[0066] Those skilled in the art can understand that the description of the above device is only an example and does not constitute a limitation on the device. It may include more or fewer components than the above description, or combine some components, or different components. For example, it may include input / output devices, network access devices, buses, etc. The so-called processor may be a central processing unit (CPU), or it may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the computer device and connects various parts of the entire computer device through various interfaces and lines.

[0067] The memory can be used to store computer programs and / or modules. The processor realizes various functions of the computer device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the mobile phone (such as audio data, phone book, etc.). In addition, the memory may include high-speed random access memory and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one magnetic disk storage device, flash device, or other volatile solid-state storage devices. If the integrated modules / units of the device are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program.

[0068] Embodiment 3:

[0069] The driverless vehicle in this embodiment includes a driverless vehicle body, and a driverless vehicle automatic turning control device as described in Embodiment 2 is further provided on the driverless vehicle body. The driverless vehicle automatic turning control device includes a processor and a memory. The memory is used to store a computer program, and the processor is used to execute the computer program. The processor is used to execute the computer program to perform the driverless vehicle automatic turning control method. During the process of the driverless vehicle driving along the required task path, when it is recognized that a turn is needed, it controls to track the task path and go straight until it reaches the turning section and then performs path tracking and turning. During the path tracking and turning process, when an obstacle is recognized, a barrier-free path in the same direction as the turning end point is searched as the target path according to the task path, and a set of target points is obtained from the searched target path. A feasible path is iteratively searched according to the obtained target points until the turning is completed, so as to obtain a path result that can not only avoid obstacles but also be closer to the driving habits of humans, thereby effectively improving the environmental adaptability, motion flexibility, and turning success rate of the driverless vehicle. The above driverless vehicle automatic turning control method is specifically referred to as shown in Embodiment 1 and will not be elaborated here one by one.

[0070] The above is only a preferred embodiment of the present invention and does not impose any form of limitation on the present invention. Although the present invention has been disclosed above with the preferred embodiment, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change, and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

Claims

1. An automatic turning control method for a driverless vehicle, characterized in that, Including: During the process of the driverless vehicle driving along the required task path, when it is recognized that a U-turn is needed, control the vehicle to track the task path straight until it reaches the U-turn section and then perform a path-tracking turn; During the path-tracking turn, when an obstacle is recognized, search for an obstacle-free path in the same direction as the U-turn end point according to the task path as the target path, and obtain a set of target points from the searched target path; Iteratively search for a feasible path based on each of the obtained target points until the U-turn is completed; When it is recognized that a U-turn is needed, the task path is further divided into a first straight section, a U-turn section, and a second straight section. The first straight section is a straight line segment between the driving starting point where it is recognized that a U-turn is needed and the starting point of the U-turn section. The U-turn section is a turning section at the U-turn intersection. The second straight section is a straight line segment between the end point of the U-turn section and the position point where the driverless vehicle completes the U-turn; When it is recognized that a U-turn is needed, control the driverless vehicle to drive straight along the first straight section to the starting point of the U-turn section; When it is recognized that the driverless vehicle reaches the starting point of the U-turn section, control the driverless vehicle to perform a path-tracking turn along the U-turn section with the minimum turning radius; When an obstacle is recognized during the path-tracking turn, control the driverless vehicle to enter the path search state to search for the target path according to the second straight section; Searching for the target path according to the second straight section includes: using the second straight section as a reference line, generating a set of parallel reference lines by translating left and right in the lane, and selecting an obstacle-free path closest to the second straight section from each of the parallel reference lines as the finally selected target path.

2. The automatic turning control method for the driverless vehicle according to claim 1, wherein Iteratively searching for a feasible path based on each of the obtained target points includes: using the search method of the hybrid A* algorithm to iteratively search for a feasible path to reach each of the target points from far to near. During the search using the hybrid A* algorithm, the cost of the current node includes the cumulative distance value from the starting point to the current node and the cumulative obstacle distance from the starting point to the current node.

3. The automatic U-turn control method for the driverless vehicle according to claim 2, wherein The cost of the current node is obtained by weighting the cumulative distance value and the cumulative obstacle distance.

4. The automatic U-turn control method for the driverless vehicle according to any one of claims 1 to 3, characterized in that, After iteratively searching for a feasible path based on each of the obtained target points, it further includes: segmenting the searched feasible path according to the movement direction of the path, and controlling the driverless vehicle to enter the path-tracking state. During the segmented path-tracking process, generate a smooth path that meets the preset constraint requirements with the segmented path as the target path, and track the smooth path generated by each segmented path until the direction of the driverless vehicle is the same as the U-turn end point to complete the U-turn or track to the end point of the last path segment.

5. The automatic U-turn control method for the driverless vehicle according to claim 4, wherein During the segmented path-tracking process, when an obstacle is recognized, control the driverless vehicle to switch to the path search state to search for a feasible path from the current position to the end point.

6. An automatic U-turn control device for an unmanned vehicle, comprising a processor and a memory, the memory is used for storing a computer program, and the processor is used for executing the computer program, characterized in that, The processor is used to execute the computer program to execute the method according to any one of claims 1 to 5.

7. An unmanned vehicle, comprising an unmanned vehicle body, characterized in that, The driverless vehicle automatic U-turn control device according to claim 6 is further provided on the driverless vehicle body.

Images