A method, apparatus, and device for trajectory-free motion planning based on unmanned aerial vehicles (UAVs)

Abstract

A method, apparatus, and medium for trajectory-free motion planning based on unmanned aerial vehicles (UAVs) include acquiring a pre-constructed sparse neighborhood graph and motion reference strategy corresponding to a UAV reference region; executing UAV motion in the sparse neighborhood graph according to the motion reference strategy; acquiring UAV reference nodes and UAV target nodes during motion; constructing a UAV state vector based on the UAV reference nodes and UAV target nodes; and performing motion control on the UAV based on the UAV state vector and a preset controller. This provides a safe and reliable motion planning method for unpredictable UAV environments, improving the autonomous motion planning level of UAVs.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and specifically to a method, apparatus, and device for trajectory-free motion planning based on unmanned aerial vehicles (UAVs). Background Technology

[0002] Warehouses are a key component of logistics companies, manufacturing enterprises, and government agencies, playing a vital role in modern logistics services such as cargo sorting, transshipment, and distribution. Especially in recent years, the "online shopping boom" has led to a rapid increase in logistics volume. However, due to improper stacking methods, manual handling, and mechanical scratches, cargo damage is severe. Damage to the appearance of goods not only affects their aesthetics but may also impact their quality, seriously affecting the consumer shopping experience, damaging the company's brand image, and increasing operating costs. According to incomplete statistics, economic losses due to cargo damage exceed 10 billion yuan annually. For ceramics and glass products, due to their material properties, the breakage rate during logistics transportation, warehousing, and manual handling exceeds 20%, resulting in direct economic losses of hundreds of millions of yuan. For large household appliances such as refrigerators and washing machines, the large size and difficulty of manual handling lead to collisions, compression, and deformation of the outer packaging. This type of outer packaging damage accounts for more than 8% of total cargo losses, resulting in economic losses of billions of yuan annually.

[0003] Traditional warehouse goods inspection methods primarily rely on manual verification. While capable of inspecting various goods, the process is complex, time-consuming, and requires a large workforce, increasing both human and financial resources. Drone inspection, an emerging intelligent inspection method, involves mounting inspection equipment on a drone platform, allowing staff to control the drone's flight for remote, non-contact inspection of multiple targets. Currently, drone-based inspection technology is increasingly being applied in aerial photography, road surveying, and large equipment inspection. Research is underway on a method that uses RFID devices on small drones to identify barcodes on warehouse goods, helping warehouse managers resolve discrepancies between inventory data and actual material storage. Compared to traditional manual inspection, this drone-based inspection method offers advantages such as ease of operation, portability, and high efficiency. Combined with various sensors, it enables diverse applications across multiple scenarios. Therefore, the use of drones in warehousing is of significant practical importance.

[0004] In the warehousing field, the main goal of motion planning for drones is to autonomously move the drone from its initial position to a target position while adhering to constraints from the environment and the drone itself. The success of motion planning algorithms largely depends on their ability to handle obstacles and dynamics. Trackless motion planning is a novel concept; this approach is not based on a trajectory. A typical method for this type of problem involves segmenting the map using connected sub-regions, then establishing a real-time planner and motion controller for the drone to drive it to the target area planned by the drone, satisfying the constraints from the target area. Current algorithms for trackless motion planning still exhibit relatively poor autonomous drone operation, frequently colliding with warehouse shelves or the ground. Therefore, more robust algorithms are needed to achieve trackless motion planning for drones. Summary of the Invention

[0005] The main technical problem solved by this invention is to provide a safe and reliable motion planning method in unpredictable environments in which drones operate, thereby improving the autonomous motion planning level of drones.

[0006] According to the first aspect, one embodiment provides a trajectory-free motion planning method based on an unmanned aerial vehicle (UAV), characterized in that it includes:

[0007] Obtain a sparse neighborhood graph and motion reference strategy corresponding to a pre-constructed UAV reference region; the motion reference strategy is generated based on the sparse neighborhood graph.

[0008] The drone's motion is executed in the sparse neighborhood graph according to the motion reference strategy, and the drone reference node and drone target node are obtained when the drone is moving. The drone state vector is constructed based on the drone reference node and drone target node.

[0009] The drone's motion is controlled based on the drone's state vector and a preset controller.

[0010] In one embodiment, obtaining the sparse neighborhood graph and motion reference strategy corresponding to the pre-constructed UAV reference region includes:

[0011] Obtain a preset reference node and the node region corresponding to the reference node. Generate a neighborhood graph based on the UAV reference region, the preset sampling points and the node region corresponding to the reference node to obtain a sparse neighborhood graph.

[0012] Calculate the edge cost corresponding to the sparse neighborhood graph, and perform path search on the sparse neighborhood graph based on the edge cost and a preset search algorithm to obtain a motion reference strategy.

[0013] In one embodiment, a sparse neighborhood graph is generated based on the UAV reference area, pre-set sampling points, and the node region corresponding to the reference node, including:

[0014] The region within the UAV reference region after removing the node region corresponding to the reference node is selected as the target region.

[0015] Determine whether the sampling point is within the target area. If the sampling point is not within the target area, delete the sampling point and re-acquire the sampling point for determination.

[0016] When the sampling point is within the target area, a first node is generated. Based on the first node and the sampling point, multiple graph expansions are performed until a preset termination condition is met. The expanded neighborhood graph is then used as a sparse neighborhood graph.

[0017] In one embodiment, a path search is performed on the sparse neighborhood graph based on the edge cost and a preset search algorithm to obtain a motion reference strategy, including:

[0018] Identify multiple nodes in the sparse neighborhood graph and the node order of the multiple nodes according to the edge cost minimization principle corresponding to the edge cost;

[0019] Based on the node order of the multiple nodes and the search algorithm, a path is constructed for the multiple nodes to obtain a motion reference strategy.

[0020] In one embodiment, constructing a UAV state vector based on the UAV reference node and the UAV target node includes:

[0021] Identify the corresponding centroid based on the UAV reference node and the UAV target node;

[0022] A first transformation matrix and a second transformation matrix are constructed based on the centroid; the first transformation matrix represents the attitude of the target coordinate system and the world coordinate system, and the second transformation matrix represents the attitude of the UAV's corresponding coordinate system and the world coordinate system.

[0023] A corresponding third transformation matrix is ​​generated using the first transformation matrix and the second transformation matrix; the third transformation matrix represents the attitude of the UAV's coordinate system relative to the target coordinate system.

[0024] The third transformation matrix is ​​used to generate the UAV state vector corresponding to the target coordinate system.

[0025] In one embodiment, motion control of the UAV based on the UAV state vector and a preset controller includes:

[0026] The UAV state vector is used as the state constraint condition of the controller, and the controller is used to control the UAV to fly to the position corresponding to the node in the motion reference strategy;

[0027] When the position corresponding to a node in the motion reference strategy meets the preset update conditions, the drone state vector corresponding to the drone is recalculated, and the drone is motion controlled according to the recalculated drone state vector until the drone flies to the preset final node.

[0028] In one embodiment, after performing the UAV's motion according to the motion reference strategy in the sparse neighborhood graph, the method further includes:

[0029] If the node of the UAV during its movement is inconsistent with the reference node of the UAV or the UAV does not move to the reference area of ​​the UAV, the final position of the UAV after its movement shall be used as the reference node.

[0030] The sparse neighborhood graph is updated based on the reference node, and the search algorithm is used to perform path search in the updated sparse neighborhood graph to obtain the updated motion reference strategy.

[0031] According to a second aspect, one embodiment provides a trajectory-free motion planning device based on an unmanned aerial vehicle (UAV), characterized in that it includes:

[0032] The strategy generation module is used to obtain a sparse neighborhood graph and motion reference strategy corresponding to a pre-constructed UAV reference region; the motion reference strategy is generated based on the sparse neighborhood graph.

[0033] The vector construction module is used to execute the movement of the UAV in the sparse neighborhood graph according to the motion reference strategy, and to obtain the UAV reference node and UAV target node when the UAV is moving, and to construct the UAV state vector based on the UAV reference node and UAV target node.

[0034] The motion control module is used to perform motion control on the UAV based on the UAV's state vector and a preset controller.

[0035] According to a third aspect, embodiments of the present invention provide an apparatus comprising:

[0036] Memory, used to store programs;

[0037] A processor for implementing the UAV-based trajectory-free motion planning method as described in any of the preceding claims by executing a program stored in the memory.

[0038] The above-described UAV-based trajectory-free motion planning method, apparatus, and medium include acquiring a pre-constructed sparse neighborhood graph and motion reference strategy corresponding to a UAV reference region; executing UAV motion in the sparse neighborhood graph according to the motion reference strategy; acquiring UAV reference nodes and UAV target nodes during motion; constructing a UAV state vector based on the UAV reference nodes and the UAV target nodes; and performing motion control on the UAV based on the UAV state vector and a preset controller. This provides a safe and reliable motion planning method in unpredictable UAV environments, improving the autonomous motion planning level of UAVs. Attached Figure Description

[0039] Figure 1 This is a flowchart of a drone-based trajectory-free motion planning method according to an embodiment of this application.

[0040] Figure 2 This is a schematic diagram of a trajectory-free motion planning process based on an unmanned aerial vehicle (UAV) according to one embodiment.

[0041] Figure 3 This is a schematic diagram of a drone-based trajectory-free motion planning process according to another embodiment.

[0042] Figure 4 This is a schematic diagram of a drone-based trajectory-free motion planning process according to another embodiment.

[0043] Figure 5 This is a schematic diagram of a drone-based trajectory-free motion planning process according to another embodiment.

[0044] Figure 6 This is a schematic diagram of a drone-based trajectory-free motion planning process according to another embodiment.

[0045] Figure 7 This is a schematic diagram of a drone-based trajectory-free motion planning process according to another embodiment.

[0046] Figure 8 This is a structural block diagram of a drone-based trajectory-free motion planning device according to an embodiment of this application. Detailed Implementation

[0047] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of this application. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to this application are not shown or described in the specification. This is to avoid obscuring the core parts of this application with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0048] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0049] The serial numbers assigned to components in this document, such as "first" and "second," are used only to distinguish the described objects and have no sequential or technical meaning. The terms "connection" and "linkage" used in this application, unless otherwise specified, include both direct and indirect connections (linkages).

[0050] In some embodiments, a sparse neighborhood graph and motion reference strategy corresponding to a pre-constructed UAV reference region are obtained. The UAV's motion is executed in the sparse neighborhood graph according to the motion reference strategy, and the UAV reference node and UAV target node are obtained during the motion. A UAV state vector is constructed based on the UAV reference node and the UAV target node. Motion control of the UAV is performed based on the UAV state vector and a preset controller. This provides a safe and reliable motion planning method in unpredictable UAV environments, improving the autonomous motion planning level of UAVs.

[0051] Please refer to Figure 1 Some embodiments of the present invention provide a trajectory-free motion planning method based on unmanned aerial vehicles (UAVs), including steps S10 to S30, which are described in detail below.

[0052] Step S10: Obtain the sparse neighborhood graph and motion reference policy corresponding to the pre-built UAV reference region; the motion reference policy is generated based on the sparse neighborhood graph.

[0053] In some embodiments, the drone reference area refers to the area where the drone will perform its flight, for example, a warehouse area in a warehousing environment.

[0054] See Figure 2 In some embodiments, step S10, which involves obtaining the sparse neighborhood graph and motion reference strategy corresponding to the pre-constructed UAV reference region, includes steps S11 to S12, which will be described in detail below.

[0055] Step S11: Obtain the preset reference node and the node region corresponding to the reference node. Generate a neighborhood graph based on the UAV reference region, the preset sampling points and the node region corresponding to the reference node to obtain a sparse neighborhood graph.

[0056] In some embodiments, the preset reference node is a rectangular node sampled from a historical time period, and the node region corresponding to the reference node refers to the rectangular region where the rectangular node is located. In fact, nodes can be of other geometric shapes besides rectangular nodes, such as squares, circles, or triangles. Rectangular nodes are used in this embodiment because they allow for the construction of a sparser graph structure.

[0057] See Figure 3 In some embodiments, step S11 generates a neighborhood graph based on the UAV reference area, pre-set sampling points, and the node area corresponding to the reference node, and obtains a sparse neighborhood graph, including steps S111 to S113, which are described in detail below.

[0058] Step S111: Select the region in the UAV reference region after removing the node region corresponding to the reference node as the target region.

[0059] In some embodiments, it is assumed that the reference area of ​​the UAV is S1 and the node area corresponding to the reference node is S2, then the target area is S1-S2.

[0060] Step S112: Determine whether the sampling point is within the target area. If the sampling point is not within the target area, delete the sampling point and re-acquire the sampling point for judgment.

[0061] In some embodiments, the sampling point is q rand Sampling point q rand This can be achieved through pre-setting.

[0062] Step S113: When the sampling point is within the target area, generate the first node, perform multiple graph expansions based on the first node and the sampling point until the preset termination condition is met, and use the expanded neighborhood graph as the sparse neighborhood graph.

[0063] In some embodiments, when sampling point q rand When within the target area, generate the first node (Node).k According to the first node k Create edges between nodes and their adjacent nodes, so in C free In \β, multiple graphical expansions are implemented until a preset termination condition is met. Wherein, C free β represents the set of all nodes in the sparse neighborhood graph, and β represents the union of all nodes in the sparse neighborhood graph.

[0064] In some embodiments, the preset termination condition is: Where m is the number of consecutive failures after the first successful expansion, and P c α and α are preset fixed parameters, specifically user parameters that affect the coverage density of sparse neighborhood graphs.

[0065] Step S12: Calculate the edge cost corresponding to the sparse neighborhood graph, and perform path search on the sparse neighborhood graph based on the edge cost and the preset search algorithm to obtain the motion reference strategy.

[0066] In some embodiments, the edge cost corresponding to the sparse neighborhood graph is calculated using a preset edge cost calculation formula, including:

[0067] The preset formula for calculating edge cost is:

[0068]

[0069] Among them, cost edge Indicates edge cost, Let A represent the intersection of the diagonals of the i-th node. i Represents the area of ​​the intersection region between two adjacent rectangles in a sparse neighborhood graph, ref i Let represent the center of the intersecting region, γ represent the preset weight, and ||·||2 represent the Euclidean distance between the two edges.

[0070] In some embodiments, the preset weight γ is the reciprocal of the area of ​​the intersecting region, typically γ = 1. Because a larger area provides a smoother motion planning trajectory for the drone, the reciprocal of the intersection area of ​​two adjacent rectangles is used as a parameter of the edge cost.

[0071] See Figure 4 In some embodiments, step S12 performs path search on the sparse neighborhood graph based on edge cost and a preset search algorithm to obtain a motion reference strategy, including steps S121 to S122, which are described in detail below.

[0072] Step S121: Identify multiple nodes and their order in the sparse neighborhood graph based on the principle of minimizing edge costs.

[0073] In some embodiments, the node order of the multiple nodes is the same as the node order through which the drone passes.

[0074] Step S122: Construct paths for multiple nodes based on their node order and search algorithm to obtain a motion reference strategy.

[0075] In some embodiments, the search algorithm is Dijkstra's search algorithm, and the path construction yields a motion reference policy P. o Among them, the motion reference strategy P o Give the order of nodes that the drone should pass through when it reaches the target location.

[0076] Step S20: Perform the UAV's motion in the sparse neighborhood graph according to the motion reference strategy, and obtain the UAV reference node and UAV target node when the UAV is moving. Construct the UAV state vector based on the UAV reference node and UAV target node.

[0077] See Figure 5 In some embodiments, step S20, after executing the UAV's motion in the sparse neighborhood graph according to the motion reference strategy, includes steps S201 to S202, which will be described in detail below.

[0078] Step S201: If the node of the UAV during movement is inconsistent with the UAV reference node or the UAV has not moved to the UAV reference area, the final position of the UAV after movement shall be used as the reference node.

[0079] Step S202: Update the sparse neighborhood graph based on the reference nodes, and use a search algorithm to perform path search in the updated sparse neighborhood graph to obtain the updated motion reference strategy.

[0080] In some embodiments, due to unpredictable effects such as noise during drone operation, the drone may appear on a different node than the drone's reference node CurrentNode, or may not appear in the drone's reference area. In both of these cases, the drone's final position is used as the reference node, and the sparse neighborhood graph is updated using this reference node. The reference node is then used as the drone's new CurrentNode, and Dijkstra's search algorithm is used to perform path search in the updated sparse neighborhood graph to obtain the updated motion reference strategy.

[0081] See Figure 6 In some embodiments, step S20, which constructs the UAV state vector based on the UAV reference node and the UAV target node, includes steps S203 to S206, which will be described in detail below.

[0082] Step S203: Identify the corresponding centroids based on the UAV reference node and the UAV target node.

[0083] In some embodiments, the drone reference node is CurrentNode, and the drone target node is NextNode. The centroid is the center of the intersection region between the node region corresponding to the drone reference node and the node region corresponding to the drone target node. That is, the centroid is the reference node in the preset edge cost calculation formula. i The centroid includes the x-coordinate of the centroid. ref and the centroid ordinate y ref ,

[0084] Step S204: Construct the first transformation matrix and the second transformation matrix based on the centroid; the first transformation matrix is ​​the attitude of the target coordinate system and the world coordinate system, and the second transformation matrix is ​​the attitude of the UAV's corresponding coordinate system and the world coordinate system.

[0085] In some embodiments, a target coordinate system is constructed with the centroid as the origin, and a world coordinate system is defined.

[0086] In some embodiments, the first transformation matrix is:

[0087]

[0088] in, Let W represent the orientation of the target coordinate system and the world coordinate system, where W is the world coordinate system, t is the target coordinate system, θ is the angle between the target coordinate system and the world coordinate system, and x is the angle between the target coordinate system and the world coordinate system. ref Let y be the x-coordinate of the centroid. ref The ordinate is the centroid.

[0089] In some embodiments, the second transformation matrix is:

[0090]

[0091] in, Let W represent the attitude of the UAV in its corresponding coordinate system and the world coordinate system, where W is the world coordinate system and r is the UAV's corresponding coordinate system. Let x be the angle between the coordinate system corresponding to the UAV and the world coordinate system, where x represents the horizontal coordinate of the UAV's position in the world coordinate system, and y represents the vertical coordinate of the UAV's position in the world coordinate system.

[0092] Step S205: Generate the corresponding third transformation matrix using the first and second transformation matrices; the third transformation matrix represents the attitude of the UAV's coordinate system relative to the target coordinate system.

[0093] In some embodiments, a first transformation matrix is ​​used. Second transformation matrix Generate the corresponding third transformation matrix

[0094] Step S206: Generate the UAV state vector corresponding to the target coordinate system based on the third transformation matrix.

[0095] In some embodiments, a third transformation matrix is ​​used. Calculate the UAV position vector q relative to the target coordinate system. tr =[x tr ,y tr ,φ tr ] T The position vector of the UAV is substituted into the state vector of the UAV and then into the preset controller, so that the optimal finite time domain input sequence that satisfies the constraints can be obtained, thereby navigating the UAV to the origin of the target coordinate system.

[0096] Step S30: Perform motion control on the UAV based on the UAV state vector and the preset controller.

[0097] See Figure 7 In some embodiments, step S30, which performs motion control on the UAV based on the UAV state vector and a preset controller, includes steps S301 to S302, which will be described in detail below.

[0098] Step S301: Use the UAV state vector as the state constraint condition of the controller, and use the controller to control the UAV to fly to the position corresponding to the node in the motion reference strategy.

[0099] In some embodiments, the controller is a model predictive control (MPC) controller.

[0100] Step S302: When the position corresponding to the node in the motion reference strategy meets the preset update conditions, recalculate the UAV state vector corresponding to the UAV, and perform motion control on the UAV according to the recalculated UAV state vector until the UAV flies to the preset final node.

[0101] In some embodiments, the preset update condition is a new rectangular region in the sparse neighborhood graph. When the UAV enters each new rectangular region, the state constraints from the rectangular region are recalculated. This process is executed recursively until the UAV flies to the preset final node.

[0102] The above-described UAV-based trajectory-free motion planning method, apparatus, and medium include: acquiring a pre-constructed sparse neighborhood graph and motion reference strategy corresponding to a UAV reference region; executing UAV motion in the sparse neighborhood graph according to the motion reference strategy; acquiring UAV reference nodes and UAV target nodes during motion; constructing a UAV state vector based on the UAV reference nodes and UAV target nodes; and performing motion control on the UAV based on the UAV state vector and a preset controller. This provides a safe and reliable motion planning method for unpredictable UAV environments, improving the autonomous motion planning level of UAVs.

[0103] Please refer to Figure 8 Some embodiments provide a trajectory-free motion planning device based on unmanned aerial vehicles, including a strategy generation module 10, a vector construction module 20, and a motion control module 30, which are described in detail below.

[0104] The strategy generation module 10 is used to obtain the sparse neighborhood graph and motion reference strategy corresponding to the pre-built UAV reference area; the motion reference strategy is generated based on the sparse neighborhood graph.

[0105] In some embodiments, the drone reference area refers to the area where the drone will perform its flight, for example, a warehouse area in a warehousing environment.

[0106] Please return to the previous page. Figure 2 In some embodiments, the policy generation module 10 is used to obtain the sparse neighborhood graph and motion reference policy corresponding to the pre-built UAV reference area, including steps S11 to S12, which are described in detail below.

[0107] Step S11: Obtain the preset reference node and the node region corresponding to the reference node. Generate a neighborhood graph based on the UAV reference region, the preset sampling points and the node region corresponding to the reference node to obtain a sparse neighborhood graph.

[0108] In some embodiments, the preset reference node is a rectangular node sampled from a historical time period, and the node region corresponding to the reference node refers to the rectangular region where the rectangular node is located. In fact, nodes can be of other geometric shapes besides rectangular nodes, such as squares, circles, or triangles. Rectangular nodes are used in this embodiment because they allow for the construction of a sparser graph structure.

[0109] Please return to the previous page. Figure 3 In some embodiments, the strategy generation module 10 is used to generate a neighborhood graph based on the UAV reference area, pre-set sampling points and the node area corresponding to the reference node, and to obtain a sparse neighborhood graph, including steps S111 to S113, which are described in detail below.

[0110] Step S111: Select the region in the UAV reference region after removing the node region corresponding to the reference node as the target region.

[0111] In some embodiments, it is assumed that the reference area of ​​the UAV is S1 and the node area corresponding to the reference node is S2, then the target area is S1-S2.

[0112] Step S112: Determine whether the sampling point is within the target area. If the sampling point is not within the target area, delete the sampling point and re-acquire the sampling point for judgment.

[0113] In some embodiments, the sampling point is q rand Sampling point q rand This can be achieved through pre-setting.

[0114] Step S113: When the sampling point is within the target area, generate the first node, perform multiple graph expansions based on the first node and the sampling point until the preset termination condition is met, and use the expanded neighborhood graph as the sparse neighborhood graph.

[0115] In some embodiments, when sampling point q rand When within the target area, generate the first node (Node). k According to the first node k Create edges between nodes and their adjacent nodes, so in C free In \β, multiple graphical expansions are implemented until a preset termination condition is met. Wherein, C free β represents the set of all nodes in the sparse neighborhood graph, and β represents the union of all nodes in the sparse neighborhood graph.

[0116] In some embodiments, the preset termination condition is: Where m is the number of consecutive failures after the first successful expansion, and P c α and α are preset fixed parameters, specifically user parameters that affect the coverage density of sparse neighborhood graphs.

[0117] Step S12: Calculate the edge cost corresponding to the sparse neighborhood graph, and perform path search on the sparse neighborhood graph based on the edge cost and the preset search algorithm to obtain the motion reference strategy.

[0118] In some embodiments, the edge cost corresponding to the sparse neighborhood graph is calculated using a preset edge cost calculation formula, including:

[0119] The preset formula for calculating edge cost is:

[0120]

[0121] Among them, cost edge Indicates edge cost, Let A represent the intersection of the diagonals of the i-th node. i Represents the area of ​​the intersection region between two adjacent rectangles in a sparse neighborhood graph, ref i Let represent the center of the intersecting region, γ represent the preset weight, and ||·||2 represent the Euclidean distance between the two edges.

[0122] In some embodiments, the preset weight γ is the reciprocal of the area of ​​the intersecting region, typically γ = 1. Because a larger area provides a smoother motion planning trajectory for the drone, the reciprocal of the intersection area of ​​two adjacent rectangles is used as a parameter of the edge cost.

[0123] Please return to the previous page. Figure 4 In some embodiments, the strategy generation module 10 is used to perform path search on the sparse neighborhood graph according to the edge cost and the preset search algorithm to obtain the motion reference strategy, including steps S121 to S122, which are described in detail below.

[0124] Step S121: Identify multiple nodes and their order in the sparse neighborhood graph based on the principle of minimizing edge costs.

[0125] In some embodiments, the node order of the multiple nodes is the same as the node order through which the drone passes.

[0126] Step S122: Construct paths for multiple nodes based on their node order and search algorithm to obtain a motion reference strategy.

[0127] In some embodiments, the search algorithm is Dijkstra's search algorithm, and the path construction yields a motion reference policy P. o Among them, the motion reference strategy P o Give the order of nodes that the drone should pass through when it reaches the target location.

[0128] The vector construction module 20 is used to execute the motion of the UAV in the sparse neighborhood graph according to the motion reference strategy, and to obtain the UAV reference node and UAV target node when the UAV is in motion, and to construct the UAV state vector based on the UAV reference node and UAV target node.

[0129] Please return to the previous page. Figure 5 In some embodiments, the vector construction module 20 is used to perform the motion of the UAV in the sparse neighborhood graph according to the motion reference strategy, and then includes steps S201 to S202, which are described in detail below.

[0130] Step S201: If the node of the UAV during movement is inconsistent with the UAV reference node or the UAV has not moved to the UAV reference area, the final position of the UAV after movement shall be used as the reference node.

[0131] Step S202: Update the sparse neighborhood graph based on the reference nodes, and use a search algorithm to perform path search in the updated sparse neighborhood graph to obtain the updated motion reference strategy.

[0132] In some embodiments, due to unpredictable effects such as noise during drone operation, the drone may appear on a different node than the drone's reference node CurrentNode, or may not appear in the drone's reference area. In both of these cases, the drone's final position is used as the reference node, and the sparse neighborhood graph is updated using this reference node. The reference node is then used as the drone's new CurrentNode, and Dijkstra's search algorithm is used to perform path search in the updated sparse neighborhood graph to obtain the updated motion reference strategy.

[0133] Please return to the previous page. Figure 6 In some embodiments, the vector construction module 20 is used to construct the UAV state vector based on the UAV reference node and the UAV target node, including steps S203 to S206, which are described in detail below.

[0134] Step S203: Identify the corresponding centroids based on the UAV reference node and the UAV target node.

[0135] In some embodiments, the drone reference node is CurrentNode, and the drone target node is NextNode. The centroid is the center of the intersection region between the node region corresponding to the drone reference node and the node region corresponding to the drone target node. That is, the centroid is the reference node in the preset edge cost calculation formula. i The centroid includes the x-coordinate of the centroid. ref and the centroid ordinate y ref ,

[0136] Step S204: Construct the first transformation matrix and the second transformation matrix based on the centroid; the first transformation matrix is ​​the attitude of the target coordinate system and the world coordinate system, and the second transformation matrix is ​​the attitude of the UAV's corresponding coordinate system and the world coordinate system.

[0137] In some embodiments, a target coordinate system is constructed with the centroid as the origin, and a world coordinate system is defined.

[0138] In some embodiments, the first transformation matrix is:

[0139]

[0140] in, Let W represent the orientation of the target coordinate system and the world coordinate system, where W is the world coordinate system, t is the target coordinate system, θ is the angle between the target coordinate system and the world coordinate system, and x is the angle between the target coordinate system and the world coordinate system.ref Let y be the x-coordinate of the centroid. ref The ordinate is the centroid.

[0141] In some embodiments, the second transformation matrix is:

[0142]

[0143] in, Let W represent the attitude of the UAV in its corresponding coordinate system and the world coordinate system, where W is the world coordinate system and r is the UAV's corresponding coordinate system. Let x be the angle between the coordinate system corresponding to the UAV and the world coordinate system, where x represents the horizontal coordinate of the UAV's position in the world coordinate system, and y represents the vertical coordinate of the UAV's position in the world coordinate system.

[0144] Step S205: Generate the corresponding third transformation matrix using the first and second transformation matrices; the third transformation matrix represents the attitude of the UAV's coordinate system relative to the target coordinate system.

[0145] In some embodiments, a first transformation matrix is ​​used. Second transformation matrix Generate the corresponding third transformation matrix

[0146] Step S206: Generate the UAV state vector corresponding to the target coordinate system based on the third transformation matrix.

[0147] In some embodiments, a third transformation matrix is ​​used. Calculate the UAV position vector q relative to the target coordinate system. tr =[x tr ,y tr ,φ tr ] T The position vector of the UAV is substituted into the state vector of the UAV and then into the preset controller, so that the optimal finite time domain input sequence that satisfies the constraints can be obtained, thereby navigating the UAV to the origin of the target coordinate system.

[0148] The motion control module 30 is used to perform motion control on the UAV based on the UAV state vector and a preset controller.

[0149] Please return to the previous page. Figure 7 In some embodiments, the motion control module 30 is used to perform motion control on the UAV based on the UAV state vector and a preset controller, including steps S301 to S302, which are described in detail below.

[0150] Step S301: Use the UAV state vector as the state constraint condition of the controller, and use the controller to control the UAV to fly to the position corresponding to the node in the motion reference strategy.

[0151] In some embodiments, the controller is a model predictive control (MPC) controller.

[0152] Step S302: When the position corresponding to the node in the motion reference strategy meets the preset update conditions, recalculate the UAV state vector corresponding to the UAV, and perform motion control on the UAV according to the recalculated UAV state vector until the UAV flies to the preset final node.

[0153] In some embodiments, the preset update condition is a new rectangular region in the sparse neighborhood graph. When the UAV enters each new rectangular region, the state constraints from the rectangular region are recalculated. This process is executed recursively until the UAV flies to the preset final node.

[0154] The above-described UAV-based trajectory-free motion planning method, apparatus, and medium include acquiring a pre-constructed sparse neighborhood graph and motion reference strategy corresponding to a UAV reference region; executing UAV motion in the sparse neighborhood graph according to the motion reference strategy; acquiring UAV reference nodes and UAV target nodes during motion; constructing a UAV state vector based on the UAV reference nodes and UAV target nodes; and performing motion control on the UAV based on the UAV state vector and a preset controller. This provides a safe and reliable motion planning method for unpredictable UAV environments, improving the autonomous motion planning level of UAVs.

[0155] Those skilled in the art will understand that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, which may include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to achieve the above functions. For example, the program can be stored in the memory of a device, and when the program in the memory is executed by the processor, all or part of the above functions can be achieved. In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the program can also be stored in a server, another computer, disk, optical disk, flash drive, or external hard drive, etc., and can be downloaded or copied to the memory of a local device, or the system of the local device can be updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be achieved.

[0156] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A trajectory-free motion planning method based on unmanned aerial vehicles (UAVs), characterized in that, include: Obtain a preset reference node and the node region corresponding to the reference node, and select the region after removing the node region corresponding to the reference node from the pre-constructed UAV reference region as the target region; Determine whether the pre-set sampling point is within the target area. If the sampling point is not within the target area, delete the sampling point and re-acquire the sampling point for judgment. When the sampling point is within the target area, a first node is generated. Based on the first node and the sampling point, multiple graph expansions are performed until a preset termination condition is met. The expanded neighborhood graph is then used as a sparse neighborhood graph. Calculate the edge cost corresponding to the sparse neighborhood graph, and perform path search on the sparse neighborhood graph based on the edge cost and a preset search algorithm to obtain a motion reference strategy; The drone's motion is executed in the sparse neighborhood graph according to the motion reference strategy, and the drone reference node and drone target node are obtained when the drone is moving. The drone state vector is constructed based on the drone reference node and drone target node. The drone's motion is controlled based on the drone's state vector and a preset controller.

2. The method as described in claim 1, characterized in that, The step of performing path search on the sparse neighborhood graph based on the edge cost and a preset search algorithm to obtain a motion reference strategy includes: Identify multiple nodes in the sparse neighborhood graph and the node order of the multiple nodes according to the edge cost minimization principle corresponding to the edge cost; Based on the node order of the multiple nodes and the search algorithm, a path is constructed for the multiple nodes to obtain a motion reference strategy.

3. The method as described in claim 1, characterized in that, The step of constructing the UAV state vector based on the UAV reference node and the UAV target node includes: Identify the corresponding centroid based on the UAV reference node and the UAV target node; A first transformation matrix and a second transformation matrix are constructed based on the centroid; the first transformation matrix represents the attitude of the target coordinate system and the world coordinate system, and the second transformation matrix represents the attitude of the UAV's corresponding coordinate system and the world coordinate system. A corresponding third transformation matrix is ​​generated using the first transformation matrix and the second transformation matrix; the third transformation matrix represents the attitude of the UAV's coordinate system relative to the target coordinate system. The third transformation matrix is ​​used to generate the UAV state vector corresponding to the target coordinate system.

4. The method as described in claim 1, characterized in that, The motion control of the UAV based on the UAV state vector and the preset controller includes: The UAV state vector is used as the state constraint condition of the controller, and the controller is used to control the UAV to fly to the position corresponding to the node in the motion reference strategy; When the position corresponding to a node in the motion reference strategy meets the preset update conditions, the drone state vector corresponding to the drone is recalculated, and the drone is motion controlled according to the recalculated drone state vector until the drone flies to the preset final node.

5. The method as described in claim 1, characterized in that, After executing the UAV's motion in the sparse neighborhood graph according to the motion reference strategy, the method further includes: If the node of the UAV during its movement is inconsistent with the reference node of the UAV or the UAV does not move to the reference area of ​​the UAV, the final position of the UAV after its movement shall be used as the reference node. The sparse neighborhood graph is updated based on the reference node, and the search algorithm is used to perform path search in the updated sparse neighborhood graph to obtain the updated motion reference strategy.

6. A trajectory-free motion planning device based on unmanned aerial vehicles (UAVs), characterized in that, include: The strategy generation module is used to obtain preset reference nodes and the node regions corresponding to the reference nodes, and select the region after removing the node regions corresponding to the reference nodes from the pre-constructed UAV reference region as the target region. Determine whether the pre-set sampling point is within the target area. If the sampling point is not within the target area, delete the sampling point and re-acquire the sampling point for judgment. When the sampling point is within the target area, a first node is generated. Based on the first node and the sampling point, multiple graph expansions are performed until a preset termination condition is met. The expanded neighborhood graph is then used as a sparse neighborhood graph. Calculate the edge cost corresponding to the sparse neighborhood graph, and perform path search on the sparse neighborhood graph based on the edge cost and a preset search algorithm to obtain a motion reference strategy; The vector construction module is used to execute the movement of the UAV in the sparse neighborhood graph according to the motion reference strategy, and to obtain the UAV reference node and UAV target node when the UAV is moving, and to construct the UAV state vector based on the UAV reference node and UAV target node. The motion control module is used to perform motion control on the UAV based on the UAV's state vector and a preset controller.

7. A trajectory-free motion planning device based on unmanned aerial vehicles (UAVs), characterized in that, include: Memory, used to store programs; A processor for implementing the method as described in any one of claims 1-5 by executing a program stored in the memory.

8. A computer-readable storage medium, characterized in that, The medium stores a program that can be executed by a processor to implement the method as described in any one of claims 1-5.

Images