An unmanned vehicle hierarchical road network path planning method and system based on air-ground cooperation

Abstract

The application provides an unmanned vehicle hierarchical road network path planning method and system based on air-ground cooperation, and belongs to the technical field of unmanned equipment path planning. The method comprises the following steps: S1, performing preliminary global planning on a road network according to a satellite map, so as to extract an ordered global road point set with an interval of n1; S2, sequentially going to each global road point by a high-altitude unmanned aerial vehicle, so as to obtain a more detailed local map, align and fuse the two layers of road networks, supplement part of road network information, and calculate and update road points; S3, taking the working range of a sensor carried by the unmanned vehicle as a planning step, considering dynamic constraints and local dynamic and static obstacle avoidance, performing local planning for the unmanned vehicle, obtaining unmanned vehicle road points, and obtaining an expected trajectory of the unmanned vehicle through smoothing; and S4, according to the position movement of the unmanned vehicle, iteratively performing steps S2 and S3 until the unmanned vehicle reaches a target point. The method improves the planning success rate.

Description

Technical Field

[0001] This invention belongs to the field of unmanned equipment path planning technology, and in particular relates to a hierarchical road network path planning method and system for unmanned vehicles based on air-ground cooperation. Background Technology

[0002] Unmanned vehicles (RVs) are highly automated transportation tools capable of autonomous navigation, monitoring, and task execution. They are complex systems integrating multiple sensors, intelligent algorithms, and communication technologies, exhibiting a high degree of task autonomy and environmental adaptability. Due to their unique autonomous operation capabilities, they have been widely applied in logistics distribution, environmental monitoring, agricultural operations, and urban traffic management; they also play a crucial role in emergency rescue situations (such as natural disasters like mudslides and earthquakes, fires, and disaster relief). As a multi-functional autonomous equipment capable of reconnaissance, transportation, and support, their unique mobility and flexibility have led to their widespread application in reconnaissance and logistical resupply missions.

[0003] A road network refers to a series of roads and their connections used for navigation and path planning of autonomous vehicles, including road geometry, traffic rules, traffic lights, and other information. Road network data can be acquired through high-precision maps, satellite remote sensing, and sensors, while road network processing involves transforming the acquired data into a format that the autonomous vehicle navigation system can understand and process.

[0004] Current autonomous vehicle planning algorithms, on a large scale, mainly consist of autonomous driving algorithms that plan using pre-obtained satellite city road network maps. Although current autonomous driving algorithms consider both dynamic and static obstacles, their road network information is obtained in advance and is not updated. In special environments, pre-planned roads may be blocked or damaged, making them impassable. They cannot effectively cope with changes in terrain and road connectivity.

[0005] The SLAM algorithm for autonomous vehicles can perceive and plan the environment on its own using sensors. However, since the SLAM algorithm does not know the environmental information in advance, it is suitable for use in situations where the work area is small. If large-scale operations across blocks need to be completed, using only the SLAM algorithm will affect the work efficiency and prolong the work time. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention aims to provide a hierarchical road network path planning method and system for unmanned vehicles based on air-ground cooperation.

[0007] The first aspect of this invention discloses a hierarchical road network path planning method for unmanned vehicles based on air-ground cooperation, the method comprising:

[0008] S1, perform preliminary global road network planning based on satellite maps to extract an ordered global waypoint set with an interval of n1;

[0009] S2 utilizes high-altitude drones to acquire detailed local road network information and performs two-layer road network alignment and fusion, as well as path updates, including:

[0010] S21, the UAV flies sequentially to each global waypoint in the ordered global waypoint set, and collects and sends out detailed local maps in real time during the flight;

[0011] S22, based on semantic segmentation algorithm, processes fine local maps and extracts local road networks;

[0012] S23, align and merge the satellite map road network with the local road network;

[0013] S24, determine whether the connectivity of the merged road network has changed; if the connectivity of the road network has changed, return to step S1 to perform preliminary global road network planning again; otherwise, perform path planning based on the local road network to obtain each local waypoint, and combine it with the ordered global waypoint set to obtain the ordered waypoint set of the unmanned vehicle.

[0014] S25, Set the drone road network integration frequency t drone When the current time satisfies t≡0 (mod t) drone )hour,

[0015] Execute S22 to S24 periodically;

[0016] S3, based on the ordered waypoint set of the autonomous vehicle in the current period, performs local dynamic planning based on the sensor data of the autonomous vehicle, including:

[0017] S31, Establish a voxel grid covering the sensor's working range and construct a motion primitive library that conforms to dynamic constraints;

[0018] S32, establish the connection between voxels and paths, and dynamically update obstacle information based on sensor data, thereby continuously iterating the probability density of each voxel to calculate the probability density of the target point G using each path in the motion primitive library, and select the path corresponding to the voxel with the highest probability to execute.

[0019] S33, Set the local planning frequency t for the autonomous vehicle local When the current time t satisfies t≡0 (mod t) local When the driverless car iteratively executes step S32;

[0020] S4: The autonomous vehicle moves according to the global and local planning results, and S2, S3 and S4 are executed iteratively according to the changes in the autonomous vehicle's position until the target point is reached.

[0021] Optionally, step S1 specifically includes:

[0022] S11, Obtain the satellite map road network of the work area, and convert the path information in the satellite map road network into an undirected graph;

[0023] S12. Based on the undirected graph, a search-based path planning algorithm is used to perform initial path planning with the initial position of the autonomous vehicle as the starting point and the target position as the ending point, with a planning step size of n1, to obtain the global path.

[0024] S13, extract an ordered set of global waypoints with an interval of n1 along the global path.

[0025] Optionally, in step S12, the search-based path planning algorithm is the A* algorithm.

[0026] Optionally, in step S21, the drone's flight altitude h drone Determined based on matching the camera's field of view with the map size.

[0027] Optionally, in step S24, if the merged road network shows that the roads in the area where the waypoints are located are damaged, then it is determined that the road network connectivity has changed.

[0028] Optionally, in step S31, the motion primitive library is generated using B-spline curves to ensure compliance with the dynamic constraints of the unmanned vehicle.

[0029] Optionally, the frequency t of drone road network integration drone And the local planning frequency t of autonomous vehicles local Satisfy t drone <t local .

[0030] A second aspect of this invention discloses a hierarchical road network path planning system for unmanned vehicles based on air-ground cooperation, the system comprising:

[0031] The satellite map processing module is configured to perform preliminary global road network planning based on the satellite map to extract an ordered global waypoint set with an interval of n1.

[0032] The UAV road network fusion module is configured to acquire local fine-grained road network information using high-altitude UAVs and perform two-layer road network alignment fusion and path updating; specifically, the UAV road network fusion module is configured to perform the following operations:

[0033] S21, the UAV flies sequentially to each global waypoint in the ordered global waypoint set, and collects and sends out detailed local maps in real time during the flight;

[0034] S22, based on semantic segmentation algorithm, processes fine local maps and extracts local road networks;

[0035] S23, align and merge the satellite map road network with the local road network;

[0036] S24, determine whether the connectivity of the merged road network has changed; if the connectivity of the road network has changed, call the satellite map processing module to re-perform the preliminary global planning of the road network; otherwise, perform path planning based on the local road network to obtain each local waypoint, and combine it with the ordered global waypoint set to obtain the ordered waypoint set of the unmanned vehicle.

[0037] S25, Set the drone road network integration frequency t drone When the current time satisfies t≡0 (mod t) drone When this happens, S22 to S24 are executed periodically;

[0038] The autonomous vehicle local planning module is configured to perform local dynamic planning based on the ordered set of waypoints for the autonomous vehicle in the current period and according to the sensor data of the autonomous vehicle; specifically, the autonomous vehicle local planning module is configured to perform the following operations:

[0039] S31, Establish a voxel grid covering the sensor's working range and construct a motion primitive library that conforms to dynamic constraints;

[0040] S32, establish the connection between voxels and paths, and dynamically update obstacle information based on sensor data, thereby continuously iterating the probability density of each voxel to calculate the probability density of the target point G using each path in the motion primitive library, and select the path corresponding to the voxel with the highest probability to execute.

[0041] S33, Set the local planning frequency t for the autonomous vehicle local When the current time t satisfies t≡0 (mod t) local When the driverless car iteratively executes step S32;

[0042] The dynamic iterative control module is configured to control the unmanned vehicle to move according to the global and local planning results, and to iteratively call the UAV road network fusion module and the unmanned vehicle local planning module to perform relevant operations according to the changes in the unmanned vehicle's position, until the target point is reached.

[0043] A third aspect of this invention discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the hierarchical road network path planning method for unmanned vehicles based on air-ground cooperation described in the first aspect of this invention.

[0044] A fourth aspect of this invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the hierarchical road network path planning method for unmanned vehicles based on air-ground cooperation described in the first aspect of this invention.

[0045] In summary, the solution proposed in this invention has the following technical effects:

[0046] The invention takes into account changes, damage, and impacts of terrain in special environments. Autonomous vehicles require relatively flat, wide roads that can accommodate vehicle traffic. During mission execution, roads are easily obstructed by objects such as gravel, trees, damaged buildings, and road obstacles, affecting road connectivity. The hierarchical road network planning algorithm designed in this invention updates road information via a high-altitude drone and replans based on assessments to avoid planning failures caused by sudden road damage. This allows the autonomous vehicle to quickly respond to changes in road conditions, ensuring that all planned paths are executable.

[0047] By coordinating hierarchical road network information with planning algorithms of varying precision and frequency, planning efficiency is improved. To ensure the safety of personnel, unmanned vehicles (UAVs) typically begin their missions autonomously at the boundaries of specific areas, requiring them to traverse natural environments or urban blocks to approach the target. For operational ranges of hundreds or thousands of meters, relying solely on the UAV for mapping and planning is inefficient. Lower-precision global planning based on road information from satellite maps can guide the direction of movement for both personnel and vehicles. Further, real-time road information obtained from high-altitude reconnaissance drones prevents the globally planned route from being obstructed by terrain changes, and more precise waypoints are provided through the fusion of road network information. Finally, the trajectory is calculated based on the UAV's own sensors. The planning radius decreases sequentially from global planning using satellite map road networks to real-time planning using fused road networks, while the planning precision and replanning frequency increase sequentially. This approach better adapts to changing environmental conditions and improves the planning success rate. Compared to methods relying solely on the UAV's own mapping and planning, this method reduces the computational burden. Attached Figure Description

[0048] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0049] Figure 1 This is a flowchart of the hierarchical road network path planning algorithm for unmanned vehicles based on an unmanned air-ground cooperative system, according to an embodiment of the present invention.

[0050] Figure 2(a) shows a satellite map obtained in an embodiment of the present invention;

[0051] Figure 2(b) shows the undirected graph obtained by map processing according to an embodiment of the present invention.

[0052] Figure 2(c) is a schematic diagram of global waypoints obtained by global planning in an embodiment of the present invention;

[0053] Figure 3 This is a rendering of the global road network planning effect of a satellite map according to an embodiment of the present invention;

[0054] Figure 4(a) shows a local map obtained by a high-altitude reconnaissance UAV according to an embodiment of the present invention;

[0055] Figure 4(b) is a schematic diagram of road network information obtained by converting a top view through image processing and semantic segmentation algorithms according to an embodiment of the present invention;

[0056] Figure 4(c) is a schematic diagram of road network fusion through coordinate alignment in an embodiment of the present invention;

[0057] Figure 5(a) shows the global waypoint update map obtained in an embodiment of the present invention when global planning needs to be re-performed due to changes in road network connectivity.

[0058] Figure 5(b) is a schematic diagram of obtaining fine waypoints by performing local planning while maintaining the road network connectivity in an embodiment of the present invention;

[0059] Figure 6(a) is a schematic diagram of the voxel grid in embodiment (a) of the present invention;

[0060] Figure 6(b) is a schematic diagram of the motion primitive library of the unmanned vehicle in embodiment (b) of the present invention;

[0061] Figure 6(c) is a schematic diagram of the path and voxel correspondence and probability transmission in embodiment (c) of the present invention;

[0062] Figure 7 This is a structural diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] In dynamic and challenging emergency rescue environments, when using unmanned vehicles (UAVs) for missions, it's crucial to consider road damage caused by unforeseen accidents, rather than relying solely on pre-obtained satellite maps for global path planning. Furthermore, UAVs have limited field of view; currently, many unmanned systems are equipped with drones for high-altitude reconnaissance. These drones can supplement the UAVs with road network information, and obstacle avoidance, considering both dynamic and static obstacles, must be considered, with real-time responsiveness to environmental changes. Therefore, a path planning algorithm for UAVs in air-ground collaborative unmanned systems for emergency rescue scenarios is needed. This algorithm should consider multi-scale environmental changes under special conditions, layer the road network, and incorporate path planning algorithms of varying granularities to adapt to rapidly changing mission environments and improve planning efficiency.

[0065] Because autonomous vehicles (RVs) need to consider both dynamic and static obstacle avoidance, traditional RV path planning algorithms decompose the problem into high-level planning and local planning. High-level planning uses sensors to obtain environmental information and iteratively builds a local global map; it then performs initial path planning using a simple search-based path planning algorithm. Further, the local planner considers the RV's kinematics and dynamics, planning a smooth trajectory while performing local obstacle avoidance. However, in special environments, to reduce personnel casualties and improve mission autonomy, RVs need to approach the target from hundreds or even kilometers away and execute the task. If only the RV's own sensors are used to build the map and perform planning, too much useless environmental detail will be stored, the planning algorithm will easily get trapped in local minima, and it will struggle to cope with environmental changes in special environments, requiring the map to be rebuilt, significantly impacting work efficiency.

[0066] Therefore, under the background of air-ground collaboration, how to improve the path planning algorithm of unmanned vehicles to better utilize the advantages of air-ground collaboration and cope with the frequent changes in terrain and road connectivity in special environments has become a technical problem that needs to be solved by those skilled in the art.

[0067] The overall architecture of this invention is as follows:

[0068] S1. Obtain the satellite map corresponding to the air-ground cooperative system, convert the path into an undirected graph, and obtain the approximate path from the starting point to the target point through a search-based path planning algorithm. A series of ordered global waypoints with a step size of n1 are obtained. S2. A high-altitude UAV carrying an RGB camera sequentially visits the global coarse waypoints to obtain a more refined local map. The two road networks are aligned and merged, supplementing some road network information, and updated waypoints with a step size of n2 are calculated. S3. Using the working range of the sensors carried by the UAV as the planning step size, considering dynamic constraints and local dynamic and static obstacle avoidance, local planning is performed for the UAV, obtaining UAV waypoints with a step size of n3. After smoothing, the final expected trajectory of the UAV is obtained. S4. Based on the UAV's position movement, steps S2, S3, and S4 are iterated continuously until the UAV reaches the target point.

[0069] The following section provides a detailed description of the hierarchical road network path planning algorithm for unmanned vehicles based on air-ground cooperation, with reference to the accompanying drawings and specific embodiments.

[0070] like Figure 1 As shown, the hierarchical road network path planning algorithm for unmanned vehicles based on air-ground cooperation provided by this invention includes the following steps performed in sequence:

[0071] S1. Based on satellite maps, conduct preliminary global road network planning:

[0072] S11. Based on the determined work area, download the corresponding satellite map of the area in .osm format, which contains important road network information such as roads, as shown in Figure 2(a);

[0073] S12, process the satellite map, select the passable map information type according to the size of the unmanned vehicle, and convert the path information in the map into an undirected graph, as shown in Figure 2(b);

[0074] S13, using the A* algorithm, with the initial position of the autonomous vehicle as the starting point (denoted as p). start The target location is the endpoint (let's call it p). goal ), and perform planning with a step size of n1 and a planning threshold of The path planning is shown in Figure 2(c);

[0075] S14 yields a series of w ordered global waypoints with an interval of n1, denoted as p. i (i = 1, 2, ..., w), then the ordered waypoints U = {p} obtained by the autonomous vehicle so far. start ,p1,p2,...,p w ,p goal},like Figure 3 As shown;

[0076] S2. Use high-altitude UAVs to obtain specific information on the global path, and perform alignment and registration of the two-layer road network and update the planned path:

[0077] S21, a high-altitude drone carrying an RGB camera flies to a suitable altitude h above the work area. drone (Based on a combination of the drone's speed and the camera's working range), maintain the RGB camera's field of view for ground reconnaissance; set at an altitude of h... drone The map size that the camera can observe is H. RGB ×W RGB It should satisfy min{H RGB W RGB}>n1, as shown in Figure 4(a);

[0078] S22, the high-altitude UAV uses the waypoint set U obtained in step S14 as the UAV target point set, flies to each global waypoint in the order of the points in set U, and collects more detailed local map information in real time during the flight, and sends the map information to the UAV vehicle.

[0079] S23, perform image processing on the local map information and obtain local road network information based on semantic segmentation algorithm, as shown in Figure 4(b);

[0080] S24, the first-layer road network obtained by satellite map processing is aligned and merged with the second-layer road network obtained by high-altitude UAV processing, and the specific path information of some road networks is refined, as shown in Figure 4(c);

[0081] S25, after merging the two-layer road network, determine whether the merged road network affects the global planning: if waypoints If the terrain or roads in the area are damaged, the connectivity of the globally planned path changes, requiring a re-planning of the global path according to step S13 to obtain a new global waypoint sequence, as shown in step S14. After the re-planning, step S2 is then performed, as shown in Figure 5(a). If there is no damage to the area, the road network connectivity remains unchanged, based on the obtained global waypoint p. i With global waypoint p i+1 With detailed road network information, autonomous vehicles can plan routes with a step size of n2 (n2 < n1) and a planning threshold of [missing information]. Specific waypoints As shown in Figure 5(b);

[0082] S26, using S25, we can obtain the current ordered waypoint set U of the autonomous vehicle and update it as follows:

[0083] S27, Design of UAV road network fusion frequency t drone When the current time t satisfies t≡0 (mod t) droneWhen the drone executes steps S22 to S26, it can continuously calculate more precise waypoints for the unmanned vehicle.

[0084] S3. Referring to the Falco planner research from Carnegie Mellon University, a local planning method for autonomous vehicles considering dynamics and obstacle avoidance is constructed:

[0085] S31, Establish a voxel grid, with each voxel v i Includes direction d (the direction in which the autonomous vehicle enters the voxel) and position information x. The information contained in the voxel is based on voxel data. The mesh covers the sensor working range S of the autonomous vehicle as shown in Figure 6(a). Since the motion model of the autonomous vehicle can be obtained in advance, the possible paths that the autonomous vehicle may execute within the sensor range S can be obtained, thus forming the motion primitive library of the autonomous vehicle. Each path is generated using B-spline curves and conforms to dynamic constraints, as shown in Figure 6(b).

[0086] S32 establishes the connection between voxels and paths: when an obstacle exists at the center of a voxel, the autonomous vehicle tends to execute a collision-free trajectory in the adjacent direction, and probability transfer can be performed between adjacent voxels (the probability is initialized to the same value before planning begins); let position x be a voxel. The probability density of reaching the target point G is The probability density of each voxel will be continuously updated through iterative calculation, as shown in Figure 6(c).

[0087] S33 processes environmental data acquired online using the autonomous vehicle's sensors and labels paths obscured by obstacles, thereby iteratively influencing the probability density of each voxel. It then calculates the probability density of the target point G for each path in the autonomous vehicle's motion primitive library and selects p... G (v) Execute the highest-ranking set of paths;

[0088] S34, Design of local planning frequency t for autonomous vehicles local (t local =t drone When the current time t satisfies t≡0 (mod t) local When the vehicle executes steps S32 and S33, it will use the path corresponding to the voxel with the highest probability.

[0089] S4. The autonomous vehicle obtains its position guidance through global waypoints and updated waypoints, and uses its own sensors to dynamically avoid obstacles, continuously moving towards the target point. Based on the changes in the autonomous vehicle's position, steps S2, S3, and S4 are iterated until the autonomous vehicle reaches the target point.

[0090] The hierarchical road network path planning algorithm for unmanned vehicles based on air-ground cooperation provided by this invention has the following beneficial effects:

[0091] The impact of disasters, accidents, and special circumstances on terrain damage and road connectivity under special conditions is taken into consideration. Autonomous vehicles require relatively smooth, wide roads that can accommodate vehicle traffic. However, in special circumstances, roads can easily be blocked by gravel, trees, damaged buildings, and other obstacles, affecting road connectivity. The hierarchical road network planning algorithm designed in this invention updates road information via a high-altitude drone and replans based on assessments to avoid planning failures caused by sudden road damage. This allows autonomous vehicles to quickly respond to changes in road conditions, ensuring that all planned paths are executable.

[0092] By coordinating hierarchical road network information with planning algorithms of varying precision and frequency, planning efficiency is improved. Autonomous vehicles typically operate within a range of hundreds or even kilometers. Relying solely on the vehicle itself for mapping and planning is inefficient. This method utilizes lower-precision global planning based on road information from satellite maps to guide the vehicle's movement. Further, real-time road information obtained from high-altitude reconnaissance drones prevents the globally planned route from being obstructed by terrain changes, and more precise waypoints are provided through the fusion of road network information. Finally, the vehicle's own sensors are used to calculate the trajectory. The planning radius decreases sequentially from global planning using satellite map road networks to real-time planning using fused road networks, while the planning precision and replanning frequency increase sequentially. This approach better adapts to changing environmental conditions and improves the planning success rate. Compared to methods relying solely on the vehicle's own mapping and planning, this method reduces the computational burden.

[0093] A second aspect of this invention discloses a hierarchical road network path planning system for unmanned vehicles based on air-ground cooperation, the system comprising:

[0094] The satellite map processing module is configured to perform preliminary global road network planning based on the satellite map to extract an ordered global waypoint set with an interval of n1.

[0095] The UAV road network fusion module is configured to acquire local fine-grained road network information using high-altitude UAVs and perform two-layer road network alignment fusion and path updating; specifically, the UAV road network fusion module is configured to perform the following operations:

[0096] S21, the UAV flies sequentially to each global waypoint in the ordered global waypoint set, and collects and sends out detailed local maps in real time during the flight;

[0097] S22, based on semantic segmentation algorithm, processes fine local maps and extracts local road networks;

[0098] S23, align and merge the satellite map road network with the local road network;

[0099] S24, determine whether the connectivity of the merged road network has changed; if the connectivity of the road network has changed, call the satellite map processing module to re-perform the preliminary global planning of the road network; otherwise, perform path planning based on the local road network to obtain each local waypoint, and combine it with the ordered global waypoint set to obtain the ordered waypoint set of the unmanned vehicle.

[0100] S25, Set the drone road network integration frequency t drone When the current time satisfies t≡0 (mod t) drone When this happens, S22 to S24 are executed periodically;

[0101] The autonomous vehicle local planning module is configured to perform local dynamic planning based on the ordered set of waypoints for the autonomous vehicle in the current period and according to the sensor data of the autonomous vehicle; specifically, the autonomous vehicle local planning module is configured to perform the following operations:

[0102] S31, Establish a voxel grid covering the sensor's working range and construct a motion primitive library that conforms to dynamic constraints;

[0103] S32, establish the connection between voxels and paths, and dynamically update obstacle information based on sensor data, thereby continuously iterating the probability density of each voxel to calculate the probability density of the target point G using each path in the motion primitive library, and select the path corresponding to the voxel with the highest probability to execute.

[0104] S33, Set the local planning frequency t for the autonomous vehicle local When the current time t satisfies t≡0 (mod t) local When the driverless car iteratively executes step S32;

[0105] The dynamic iterative control module is configured to control the unmanned vehicle to move according to the global and local planning results, and to iteratively call the UAV road network fusion module and the unmanned vehicle local planning module to perform relevant operations according to the changes in the unmanned vehicle's position, until the target point is reached.

[0106] A third aspect of this invention discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the hierarchical road network path planning method for unmanned vehicles based on air-ground cooperation described in the first aspect of this invention.

[0107] Figure 7 This is a structural diagram of an electronic device according to an embodiment of the present invention, such as... Figure 7As shown, the electronic device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, Near Field Communication (NFC), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the device's casing, or an external keyboard, touchpad, or mouse.

[0108] Those skilled in the art will understand that Figure 7 The structure shown is merely a structural diagram of the part related to the technical solution of this disclosure and does not constitute a limitation on the electronic device to which the solution of this application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0109] A fourth aspect of this invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the hierarchical road network path planning method for unmanned vehicles based on air-ground cooperation described in the first aspect of this invention.

[0110] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein, and such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A hierarchical road network path planning method for unmanned vehicles based on air-ground cooperation, characterized in that, The method includes: S1, perform preliminary global road network planning based on satellite maps, with the extraction interval as... An ordered global waypoint set; Step S1 specifically includes: S11, Obtain the satellite map road network of the work area, and convert the path information in the satellite map road network into an undirected graph; S12, based on the undirected graph, a search-based path planning algorithm is used, with the initial position of the autonomous vehicle as the starting point and the target position as the ending point, and the planning step size is... Perform initial path planning to obtain the global path; S13, extracting intervals along the global path is... An ordered set of global waypoints; S2 utilizes high-altitude drones to acquire detailed local road network information and performs two-layer road network alignment and fusion, as well as path updates, including: S21, the UAV flies sequentially to each global waypoint in the ordered global waypoint set, and collects and sends out detailed local maps in real time during the flight; S22, based on semantic segmentation algorithm, processes fine local maps and extracts local road networks; S23, align and merge the satellite map road network with the local road network; S24, determine whether the connectivity of the merged road network has changed; if the connectivity has changed, return to step S1 to perform preliminary global road network planning again; otherwise, perform path planning based on the local road network to obtain local waypoints, and combine them with the ordered global waypoint set to obtain the ordered waypoint set of the unmanned vehicle; in step S24, if the merged road network shows that the roads in the area where the waypoints are located are damaged, it is determined that the connectivity of the road network has changed. S25, setting the frequency for drone road network integration When the current time satisfies At that time, S22 to S24 are executed periodically; S3, based on the ordered waypoint set of the autonomous vehicle in the current period, performs local dynamic planning based on the sensor data of the autonomous vehicle, including: S31, Establish a voxel grid covering the sensor's working range and construct a motion primitive library that conforms to dynamic constraints; S32 establishes the connection between voxels and paths, and dynamically updates obstacle information based on sensor data, thereby continuously iterating the probability density affecting each voxel to use each path in the motion primitive library to target the point. Calculate the probability density and select the path corresponding to the voxel with the highest probability to execute; S33, setting the local planning frequency for autonomous vehicles When the current time satisfy At that time, the unmanned vehicle iteratively executes step S32; S4, the autonomous vehicle moves according to the results of global and local planning; S5. Based on the change in the position of the unmanned vehicle, iteratively execute steps S2-S4 until the target point is reached.

2. The method according to claim 1, characterized in that, In step S12, the search-based path planning algorithm is as follows: algorithm.

3. The method according to claim 1, characterized in that, In step S21, the drone's flight altitude Determined based on matching the camera's field of view with the map size.

4. The method according to claim 1, characterized in that, In step S31, the motion primitive library is generated using B-spline curves to ensure compliance with the dynamic constraints of the unmanned vehicle.

5. The method according to claim 1, characterized in that, Unmanned aerial vehicle (UAV) road network fusion frequency and local planning frequency of autonomous vehicles satisfy < .

6. A hierarchical road network path planning system for unmanned vehicles based on air-ground cooperation, characterized in that, The system includes: The satellite map processing module is configured to perform preliminary global road network planning based on satellite maps, with the extraction interval as... An ordered set of global waypoints; specifically configured to perform the following operations: S11, Obtain the satellite map road network of the work area, and convert the path information in the satellite map road network into an undirected graph; S12, based on the undirected graph, a search-based path planning algorithm is used, with the initial position of the autonomous vehicle as the starting point and the target position as the ending point, and the planning step size is... Perform initial path planning to obtain the global path; S13, extracting intervals along the global path is... An ordered set of global waypoints; The UAV road network fusion module is configured to acquire local fine-grained road network information using high-altitude UAVs and perform two-layer road network alignment fusion and path updating; specifically, the UAV road network fusion module is configured to perform the following operations: S21, the UAV flies sequentially to each global waypoint in the ordered global waypoint set, and collects and sends out detailed local maps in real time during the flight; S22, based on semantic segmentation algorithm, processes fine local maps and extracts local road networks; S23, align and merge the satellite map road network with the local road network; S24, determine whether the connectivity of the merged road network has changed; if the connectivity has changed, call the satellite map processing module to re-perform the preliminary global planning of the road network; otherwise, perform path planning based on the local road network to obtain each local waypoint, and combine it with the ordered global waypoint set to obtain the ordered waypoint set of the unmanned vehicle; if the merged road network shows that the roads in the area where the waypoints are located are damaged, it is determined that the connectivity of the road network has changed. S25, setting the frequency for drone road network integration When the current time satisfies At that time, S22 to S24 are executed periodically; The autonomous vehicle local planning module is configured to perform local dynamic planning based on the ordered set of waypoints for the autonomous vehicle in the current period and according to the sensor data of the autonomous vehicle; specifically, the autonomous vehicle local planning module is configured to perform the following operations: S31, Establish a voxel grid covering the sensor's working range and construct a motion primitive library that conforms to dynamic constraints; S32 establishes the connection between voxels and paths, and dynamically updates obstacle information based on sensor data, thereby continuously iterating the probability density affecting each voxel to use each path in the motion primitive library to target the point. Calculate the probability density and select the path corresponding to the voxel with the highest probability to execute; S33, setting the local planning frequency for autonomous vehicles When the current time satisfy At that time, the unmanned vehicle iteratively executes step S32; The dynamic iterative control module is configured to control the unmanned vehicle to move according to the global and local planning results, and to iteratively call the UAV road network fusion module and the unmanned vehicle local planning module to perform relevant operations according to the changes in the unmanned vehicle's position, until the target point is reached.

7. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it implements the steps in the hierarchical road network path planning method for unmanned vehicles based on air-ground cooperation as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the hierarchical road network path planning method for unmanned vehicles based on air-ground cooperation as described in any one of claims 1 to 5.

Images