Amphibious path planning methods, devices, equipment and media for land and air robots

Abstract

This invention primarily applies to the field of intelligent robot technology. It discloses an amphibious path planning method, apparatus, device, and medium for land-air robots. First, a road network environment map containing multiple nodes is generated based on geographic and road network data, with the nodes distributed according to preset rules to form the road network. Next, the current starting point and target endpoint are determined from the road network nodes, and a land-based path from the starting point to the endpoint is generated along the node sequence. Within the land-based path, nodes other than the target endpoint are selected as takeoff points, and landing points are determined through a preset search method, generating a flight path. Using the landing point as a new starting point, the steps of generating the land-based path are repeated until the current starting point reaches the target endpoint, ultimately outputting an amphibious path containing all land-based and flight paths. This solution achieves efficient planning of air-ground collaborative paths through the coordinated optimization of land-based and flight paths, significantly improving the operational efficiency and task execution capabilities of land-air robots in complex environments.

Description

Technical Field

[0001] This invention relates to the field of intelligent robot technology, specifically to an amphibious path planning method, apparatus, equipment, and medium for land and air robots. Background Technology

[0002] With the deepening of intelligent transportation and smart city construction, amphibious robots with cross-domain mobility are demonstrating unique advantages in scenarios such as emergency rescue, logistics delivery, and inspection and monitoring. Existing research mostly treats ground robots and unmanned aerial vehicle (UAV) systems independently: ground path planning focuses on solving the optimal route under road network topology constraints, usually based on graph theory methods to calculate the shortest path or minimum time path in discrete space; aerial path planning focuses on trajectory generation under obstacle avoidance and endurance constraints in three-dimensional continuous space, using random sampling or optimization methods to achieve optimal single-system performance.

[0003] However, in complex urban environments characterized by dense road networks, interspersed no-fly zones, and dynamically changing tasks, this fragmented planning paradigm struggles to fully realize the potential of air-ground collaboration. While ground robots can achieve efficient long-distance movement using road networks, they are constrained by traffic congestion and traffic restrictions. Unmanned aerial vehicles (UAVs) possess the advantage of three-dimensional maneuverability but face strict constraints such as energy limitations, airspace control, and communication blind spots. More importantly, existing methods lack a systematic integration of complex constraints such as the heterogeneous kinematics of amphibious robots, the cost of multimodal switching, the uncertainty of time-varying environments, and the conflict between multiple objectives related to task priorities. This leads to significant contradictions between computational efficiency, solution quality, and real-time feasibility in collaborative paths. Summary of the Invention

[0004] This invention provides an amphibious path planning method, apparatus, equipment, and medium for land-air robots, which can achieve efficient optimization and feasible decision-making of air-ground cooperative paths under complex constraints, and significantly improve the operational efficiency and task execution capability of land-air robots in complex environments.

[0005] This invention provides an amphibious path planning method for land-air robots, the method comprising:

[0006] Based on geographic data and road network data, a road network environment map is generated, wherein the road network environment map includes multiple nodes, and each node is distributed according to preset rules to form a road network; Determine the current starting point and the target destination from multiple nodes of the road network; A land route from the current starting point to the target destination is generated along a sequence of nodes arranged in a predetermined order in the road network; Using target nodes other than the target endpoint in the land route as take-off points, and determining landing points according to a preset search method, a flight path from the take-off point to the landing point is generated; Using the landing point as the current starting point, return to the step of generating a land route from the current starting point to the target endpoint by following the node sequence arranged in a set order in the road network; When the current starting point is located at the target endpoint, the output includes an amphibious path that includes all the land paths and all the flight paths.

[0007] Optionally, before determining the takeoff point, the method further includes: Calculate the terrain detour degree for each location on the land route, wherein the terrain detour degree is used to quantify the degree of detour of the path from the location along the road network to the next node; When the terrain detour exceeds a preset threshold, a target node is determined in the land travel path.

[0008] Optionally, the specific methods for determining the target node in the land travel path include: The direction from the current starting point to the target endpoint is used as the reference direction; The land route is divided into multiple segments according to a preset step length, wherein each preset step length includes at least one node. After determining the offset of each road segment, the cumulative heading offset is obtained by accumulating the offsets, where the offset is the offset of the heading angle of the road segment relative to the reference direction; When the cumulative heading offset exceeds the preset offset threshold for the first time, the node in the current step size is taken as the target node.

[0009] Optionally, the specific methods for determining the landing point according to the preset search method include: Using the takeoff point as a reference point, a sector-shaped area is established, wherein the centerline of the sector-shaped area points towards the target endpoint, and the sector-shaped area has a preset sector angle and search radius; If no node is present within the sector area, the search radius is increased, the angle range of the sector angle is adjusted, or the generation of the flight path is stopped. If at least one of the nodes exists within the sector area, then each of the nodes located within the sector area is considered a potential landing point; Based on the flight cost from the starting point to each of the potential landing points, a target landing point is selected from a plurality of potential landing points.

[0010] Optionally, the amphibious path planning method for the land-air robot further includes parameter optimization operations on key parameters, wherein the key parameters are the preset threshold, the search radius, or the sector angle, and the implementation methods of the parameter optimization operations include: Combine the key parameters into a decision variable set; A multi-objective optimization model is constructed with the goal of reducing travel time and energy consumption. The decision variable set is co-optimized using the multi-objective optimization model to obtain a Pareto optimal solution set, where each Pareto solution contains a set of optimized decision variable values.

[0011] Optionally, the specific methods for using the multi-objective optimization model to collaboratively optimize the decision variable set include: Generate a weight value for each variable in the decision variable group, wherein the weight value is used to characterize the degree of influence of the variable on the optimization objective; The decision variable group is iteratively optimized multiple times, and after each iteration, the optimized variables with weight values ​​higher than the preset level value are targeted for optimization. When performing a mutation operation on the target variable during each iteration of optimization, a Gaussian perturbation is applied to the target variable, wherein the strength of the Gaussian perturbation is proportional to the weight value corresponding to the target variable and the range of parameter values ​​of the target variable, and the strength of the Gaussian perturbation decreases as the number of iterations increases.

[0012] Optionally, in each iteration of optimization, the amphibious path planning method for the land-air robot further includes: Based on a preset reference vector, the target variable that best approximates the Pareto front direction indicated by the reference vector is determined, forming a first candidate solution set including each target variable; Generate a normalized hyperplane based on the value of each variable in the current iteration of optimization; Each variable in the first candidate solution set is mapped to the normalized hyperplane for comparison, in order to select the variables for the next iteration of optimization.

[0013] The present invention also provides an amphibious path planning device for land and air robots, the device comprising: The mapping module is used to generate a road network environment map based on geographic data and road network data. The road network environment map includes multiple nodes, and each node is distributed according to preset rules to form a road network. The setting module is used to determine the current starting point and the target endpoint from multiple nodes of the road network; The planning module is used to generate a land route from the current starting point to the target endpoint along a sequence of nodes arranged in a preset order in the road network; using target nodes other than the target endpoint in the land route as take-off points and determining landing points according to a preset search method, a flight path from the take-off point to the landing point is generated; using the landing point as the current starting point, the module returns to the step of generating a land route from the current starting point to the target endpoint along a sequence of nodes arranged in a preset order in the road network. The output module is used to output an amphibious path that includes all the land paths and all the flight paths when the current starting point is located at the target endpoint.

[0014] The present invention also provides an electronic device, the electronic device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the amphibious path planning method for land and air robots as described in any of the preceding claims.

[0015] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the amphibious path planning method for land and air robots as described in any of the preceding claims.

[0016] The present invention has at least the following beneficial effects: This technical solution achieves efficient optimization and feasible decision-making for air-ground cooperative paths under complex constraints by organically combining land-based and flight paths. First, a road network environment map generated based on geographic and road network data provides a precise geographic foundation for path planning, ensuring its feasibility and adaptability. Second, through alternating planning of land-based and flight paths, the robot can flexibly switch between modes of movement in complex environments, effectively avoiding obstacles and shortening path length. The generation of land-based paths follows a node sequence, ensuring the orderliness and coherence of the path; the flight path determines takeoff and landing points through a preset search method, further optimizing the path and expanding the robot's operational range. This amphibious path planning method significantly improves the operational efficiency and task execution capabilities of land-air robots in complex environments, enabling them to efficiently complete tasks even when facing complex terrain and dynamic constraints. Attached Figure Description

[0017] The accompanying drawings are provided to further understand the technical solutions of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the technical solutions of the present invention, and do not constitute a limitation on the technical solutions of the present invention.

[0018] Figure 1 This is a flowchart illustrating the steps of an amphibious path planning method for land and air robots. Figure 2This is a schematic diagram illustrating the principle of calculating cumulative heading offset in an amphibious path planning method for land and air robots. Figure 3 This is a schematic diagram illustrating the principle of searching for landing points within a sector-shaped region in an amphibious path planning method for land-air robots. Figure 4 This is a flowchart illustrating the program implementation of an amphibious path planning method for land and air robots. Figure 5 This is a road network environment map output in one embodiment, which includes land routes and flight routes; Figure 6 This is a schematic diagram of the results of multi-objective amphibious planning in the Wenchuan area (canyon) in one embodiment; Figure 7 This is a schematic diagram of the results of multi-objective amphibious planning in the Tingri region (high altitude) in one embodiment; Figure 8 This is a schematic diagram of the results of multi-objective amphibious planning in the Dalingshan area (low altitude) in one embodiment; Figure 9 This is a schematic diagram of an amphibious path planning device for a land-air robot. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] Before describing this technical solution, the key terms are explained as follows: Land-air robots: composite robot platforms with both ground driving and air flight modes, typically in the form of a composite structure of a wheeled / tracked mobile chassis and a multi-rotor aircraft.

[0021] Road network constraints: Ground modes must travel along road network nodes and edges. The road network consists of a series of passable road network nodes and their connecting edges.

[0022] DEM: or Digital Elevation Model, is a three-dimensional elevation map used to represent the function / raster data of terrain elevation as it changes with plane coordinates.

[0023] VTOL (Vertical Takeoff and Landing Strategy): The air mode adopts a three-stage flight path of "vertical takeoff, horizontal cruise, and vertical landing" to reduce the obstacle avoidance / control risks caused by low-altitude maneuvering.

[0024] Amphibious route: A global route sequence consisting of several ground routes and several air routes interspersed and spliced ​​together.

[0025] Detour Degree (Cumulative Heading Angle Deviation): A cumulative measure of the heading angle offset over the next N steps of the ground path, used to characterize "potential long-distance detours caused by continuous yaw".

[0026] Heuristic sector search: Enumerate road network nodes within a polar coordinate sector (given search radius and angle range) and select candidate landing points that satisfy prior direction and distance constraints.

[0027] Ta-MaOEA: A mission-aided ideal hyperplane multi-objective evolutionary algorithm used to optimize key parameters of amphibious planning for multiple objectives (time / energy consumption) and output Pareto solutions.

[0028] In this technical field, existing path planning research mainly focuses on the following areas: Ground-based searches primarily utilize road networks or grids; aerial searches often employ 3D path planning or simplified straight-line planning. Existing work has explored mode switching based on rules or hierarchical structures, such as triggering takeoff when the slope is too steep or the cost of overcoming obstacles is too high, and landing in a nearby feasible area; Multi-objective evolutionary algorithms (NSGA-III, RVEA, etc.) that consider time, energy consumption, and risk are used to generate Pareto solution sets.

[0029] However, most amphibious land and air plans do not incorporate "road network constraints (ground accessibility)" as a strong constraint into the switching decision. At the same time, the mode switching point can be located at any node of the road network. If the take-off and landing sequences are directly searched together, it will result in too many combinations.

[0030] The researchers of this application also found that some methods allow landing in non-road network areas or generate ground segments that cannot be connected to real roads when switching between air and ground, making it difficult to implement the path; take-off / landing points can be selected anywhere in the road network, and joint search will cause the number of feasible solutions to increase exponentially with the number of nodes, making the computational load unbearable; time and energy consumption objectives are inherently contradictory, and under the constraints of road network and terrain coupling, the solution space is high-dimensional, non-convex and non-uniformly distributed, and traditional algorithms are prone to slow convergence or insufficient solution set coverage; taking off only by slope / obstacle triggering cannot effectively identify the situation of "long detour caused by continuous yaw", which is prone to invalid flight or missing key flight segments.

[0031] To address the aforementioned technical problems, this technical solution proposes an amphibious path planning method, apparatus, equipment, and medium for land-air robots. This method enables efficient optimization and feasible decision-making for air-ground cooperative paths under complex constraints, significantly improving the operational efficiency and task execution capabilities of land-air robots in complex environments. The following are various embodiments of this technical solution.

[0032] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating the steps of an amphibious path planning method for land and air robots.

[0033] This embodiment provides an amphibious path planning method for a land-air robot, including: S101. Based on geographic data and road network data, generate a road network environment map, wherein the road network environment map includes multiple nodes, and each node is distributed according to preset rules to form a road network.

[0034] S102. Determine the current starting point and the target endpoint from multiple nodes in the road network.

[0035] S103. Generate a land route from the current starting point to the target endpoint by following the node sequence arranged in a set order in the road network.

[0036] S104. Using target nodes other than the target endpoint in the land route as take-off points, and determining the landing point according to the preset search method, a flight path from the take-off point to the landing point is generated.

[0037] S105. Using the landing point as the current starting point, return to step S103.

[0038] S106. When the current starting point is located at the target endpoint, output the amphibious path, which includes all land paths and all flight paths.

[0039] It should be noted that the number of nodes in the road network environment map is obtained by sampling the road network environment map with a set precision. For example, sampling is performed at certain intervals (sampling precision) on a 100-meter-long road, such as one sampling point every five meters. This road section includes 21 nodes. When traveling by land, the virtual trajectory between one node and another is predicted and used as the land travel path.

[0040] Understandably, this embodiment achieves efficient optimization and feasible decision-making for air-ground cooperative paths under complex constraints by organically combining land-based and flight paths. First, the road network environment map generated based on geographic and road network data provides a precise geographic basis for path planning, ensuring its feasibility and adaptability. Second, through alternating planning of land-based and flight paths, the robot can flexibly switch between modes of movement in complex environments, effectively avoiding obstacles and shortening path length. The generation of land-based paths follows a node sequence, ensuring the orderliness and coherence of the path; the flight path determines takeoff and landing points through a preset search method, further optimizing the path and expanding the robot's operational range. This amphibious path planning method can significantly improve the operational efficiency and task execution capability of land-air robots in complex environments, enabling them to efficiently complete tasks even when facing complex terrain and dynamic constraints.

[0041] In some embodiments, the road network environment map consists of a topographic image (DEM) and the road network. The DEM is used to calculate slope and height difference; the road network consists of a three-dimensional coordinate sequence of nodes and edges, serving as the feasible ground region.

[0042] In some embodiments, the land travel route can be planned using a land planning unit, employing an improved road network-oriented approach. Search for the optimal ground path on the road network node map.

[0043] Specifically, input the starting point, ending point, DEM and road network data, and set the planning parameters. Then, call the land planning unit to execute A on the road network map. Search for and obtain the pre-planned ground path from the starting point to the destination, and archive it.

[0044] Specifically, the planning parameters refer to parameters such as the pre-planning step size N, detour threshold θ_th, spatial search radius R, spatial search angle α, and heuristic weight coefficient w.

[0045] In some embodiments, the flight path can be planned using an air route planning unit, generating the air path using a VTOL (Vertical Takeoff and Landing) strategy. The flight path typically consists of three segments: vertical takeoff, horizontal cruise, and vertical landing. First, the aircraft takes off vertically to a safe altitude, then cruises horizontally, and finally lands vertically to the target node.

[0046] In some embodiments, the total travel time can be calculated using a time model. For example, segmented time models of "acceleration-uniform speed-deceleration" can be established for both ground and air travel to estimate the time taken for any segment of motion.

[0047] Total energy consumption can be calculated using an energy consumption model. For example, the ground component can be calculated using a power model for mobile robots / electric vehicles, taking into account power output and motor heat loss; the air component can be calculated using a quadcopter / VTOL energy consumption model, with ascent, cruise, and descent calculated separately and then summed.

[0048] In some embodiments, an amphibious path planning method for a land-air robot further includes, prior to determining the takeoff point: Calculate the terrain detour degree for each location on the land route, where the terrain detour degree is used to quantify the detour of the path from that location to the next node along the road network; when the terrain detour degree exceeds a preset threshold, the target node is determined in the land route.

[0049] In some embodiments, determining whether the terrain detour exceeds a preset threshold can be achieved through the following steps: Calculate the terrain detour of the ground path; If the terrain detour is less than or equal to the preset threshold θ_th, then proceed along the ground path and save that ground path segment; If the terrain detour is greater than the preset threshold θ_th, then attempt to generate a flight path.

[0050] Furthermore, the takeoff point is determined by the location with the maximum detour, and the current position is used as the origin of the polar coordinates to enumerate the road network nodes within the sector range (α, R) to form a set of candidate landing points; Choose the nearest and most directional road network node as the landing point; The air route planning unit is invoked to generate the flight trajectory, while the ground path from the current point to the landing point is planned as a reference. Compare the combined time and energy costs of air and ground-based options (or weight them according to mission preferences), retain the better option, and incorporate it into the global path.

[0051] Optionally, the amphibious route sequence of the hybrid scheme, including the flight path, can be obtained and the travel time and energy consumption can be calculated.

[0052] Understandably, this embodiment accurately identifies complex terrain areas by dividing candidate ground segments and calculating terrain detour rates. For segments with detour rates exceeding a threshold, a fan-shaped area is defined with the starting point as the center, the optimal landing point is selected, the sum of flight and ground costs is calculated, and compared with the pure ground cost to generate a better hybrid solution. This improvement significantly enhances the adaptability and economy of path planning in complex terrain, further enhances the operational efficiency and task execution capabilities of air-land robots in complex environments, and optimizes the decision-making process for air-ground cooperative paths.

[0053] Please refer to Figure 2 , Figure 2 This is a schematic diagram illustrating the principle of calculating cumulative heading offset in an amphibious path planning method for land and air robots. As the reference direction vector, This is the direction vector of travel. Let N be the deviation, where N equals 1, 2, 3...N.

[0054] In some embodiments, the specific methods for determining the target node in the land travel path include: The direction from the current starting point to the target endpoint is used as the reference direction; the land path is divided into multiple segments according to the preset step size, where each preset step size includes at least one node; after determining the offset of each segment, the cumulative heading offset is obtained, where the offset is the offset of the heading angle of the segment relative to the reference direction; when the cumulative heading offset exceeds the preset offset threshold for the first time, the node in the current step size is taken as the target node.

[0055] It should be noted that the detour degree is used to identify the situation where "the target seems to be not far away, but the direction continues to deviate, resulting in a long detour". It is implemented as follows: taking the direction of the line connecting the current position to the target position as a reference, calculate the heading angle offset of the road segment corresponding to each of the next N preset steps, and accumulate the heading angle offset.

[0056] When the cumulative offset exceeds the preset threshold θ_th, it indicates that there may be a ridge / canyon ahead that causes continuous detours. This triggers the search for a takeoff point and sets the path node with the largest cumulative offset as the takeoff point to maximize the reduction effect of flight on detours.

[0057] Understandably, this embodiment accurately identifies detour areas caused by continuous directional deviations by calculating the offset of road segments and the cumulative heading offset, and uses the node with the largest cumulative offset as the takeoff point. This improvement makes flight path planning more scientific and reasonable, effectively reducing unnecessary detours and optimizing path length and energy consumption. At the same time, combined with the original landing point optimization methods, it further enhances the adaptability and flexibility of path planning under complex terrain and constraints, significantly improving the operational efficiency and task execution capability of land-air robots in complex environments, and optimizing the overall performance of air-ground cooperative paths.

[0058] Understandably, directly searching for takeoff and landing sequences at any node in the road network would result in exponential complexity. However, the above implementation uses a combination strategy of "pre-planning and archiving + detour triggering + sector landing point" to limit the switching decision to a small number of key nodes. This reduces the number of branches from a large number of node branches to a small combination of "ground advancement branches + air branches", thereby significantly reducing the complexity of the global search and improving feasibility.

[0059] Please refer to Figure 3 , Figure 3 This diagram illustrates the principle of searching for landing points within a sector-shaped region in an amphibious path planning method for land-air robots. The diagram shows the sector range (radius R, angle α) with the current position as the origin of polar coordinates and the process of selecting candidate road network nodes.

[0060] In some embodiments, the specific method for determining the landing point according to a preset search method includes: Using the takeoff point as a reference point, a fan-shaped region is established, where the centerline of the fan-shaped region points towards the target endpoint. The fan-shaped region has a preset fan angle and search radius. If there are no nodes in the fan-shaped region, the search radius is increased, the angle range of the fan angle is adjusted, or flight path generation is stopped. If there are at least one node in the fan-shaped region, each node located in the fan-shaped region is taken as a potential landing point. Based on the flight cost from the starting point to each potential landing point, the target landing point is selected from multiple potential landing points.

[0061] It should be noted that the landing point must be a node in the road network.

[0062] In some embodiments, the step of searching for landing points within a sector-shaped region is as follows: Establish a polar coordinate search space with the current position as the origin; Set the sector angle range α and radius R, and limit the search direction to prioritize the target point; Enumerate all road network nodes within the sector as candidate landing points; Calculate the distance from the current location to each candidate point, and select the candidate point with the shortest distance (or the best overall cost) as the landing point; If there are no available nodes in the sector, then increase R or adjust α, or revert to the ground scheme.

[0063] Understandably, this embodiment, by dynamically adjusting the search radius and angle range of the sector region in the polar coordinate search space, can more flexibly adapt to the node distribution in complex environments. When there are no nodes within the sector region, the search strategy can be dynamically adjusted or replacement can be stopped to avoid ineffective planning; when multiple potential landing points exist, the optimal target landing point is selected based on flight cost, ensuring the efficiency and economy of path planning. This improvement significantly enhances the flexibility and adaptability of path planning, further improving the operational efficiency and task execution capability of land-air robots in complex environments, and optimizing the decision-making accuracy of air-ground cooperative paths.

[0064] In some embodiments, an amphibious path planning method for a land-air robot further includes parameter optimization operations on key parameters, such as a preset threshold, a preset radius, and the sector angle of a sector region. The implementation methods for the parameter optimization operations include: Multiple key parameters are combined into a set of decision variables; a multi-objective optimization model is constructed with the goal of reducing travel time and energy consumption; the multi-objective optimization model is used to perform collaborative optimization on the set of decision variables to obtain a Pareto optimal solution set, where each Pareto solution contains a set of optimized decision variable values.

[0065] It should be noted that, to obtain a high-quality time / energy tradeoff under different terrains, this embodiment introduces Ta-MaOEA to perform multi-objective optimization of key parameters. Examples of optimization variables include: spatial search angle α, spatial search radius R, detour threshold θ_th, preset step size N, and heuristic weight coefficient w. The algorithm uses (time and energy consumption) as its dual objectives (which can also be extended to more objectives), iteratively generating a Pareto optimal parameter set, thereby outputting the optimal path set.

[0066] It can be understood that, based on different optimization variables (airspace search angle α, airspace search radius R, detour threshold θ_th, preset step size N, and heuristic weight coefficient w), multiple different amphibious paths are generated between the starting point and the ending point (these can be amphibious paths with only land routes or amphibious paths containing both land and flight routes). Each combination of optimization variables corresponds to one amphibious path, and each amphibious path is collected into a path set. By outputting the optimized combination of optimization variables (i.e., the decision variable set) through the multi-objective optimization model, the corresponding amphibious path can be found from the path set.

[0067] Understandably, this embodiment combines key parameters such as preset thresholds, preset radii, and sector angles into a set of decision variables and constructs a multi-objective optimization model to achieve Pareto optimal solutions by collaboratively optimizing the process to reduce travel time and energy consumption. This improvement enables path planning to dynamically adjust key parameters to adapt to different task requirements and environmental conditions, further optimizing the combination of flight and land-based paths. The optimized parameter combination enhances the flexibility and adaptability of path planning, significantly improving the operational efficiency and task execution capabilities of air-land robots in complex environments, and strengthening the overall performance and reliability of air-ground collaborative path planning.

[0068] In some embodiments, the specific methods for using a multi-objective optimization model to collaboratively optimize a set of decision variables include: Generate a weight value for each variable in the decision variable group, where the weight value is used to characterize the degree of influence of the variable on the optimization objective; perform multiple iterations of optimization on the decision variable group, and after each iteration, perform targeted optimization on the optimized variables whose weight values ​​are higher than the preset level value; when performing a mutation operation on the target variable during each iteration, apply a Gaussian perturbation to the target variable, where the strength of the Gaussian perturbation is proportional to the weight value of the target variable and the range of parameter values ​​of the target variable, and the strength of the Gaussian perturbation decreases as the number of iterations increases.

[0069] In this embodiment, to address the issue of varying target sensitivity among different parameters in amphibious planning, Ta MaOEA introduces weight vectors to exert stronger evolutionary pressure on key parameters.

[0070] In some embodiments, an elite-guided crossover mechanism is employed to achieve targeted optimization. Specifically, after standard crossover generates offspring, high-quality solutions are selected from the elite archive, and high-contribution parameters are fine-tuned according to weight values ​​to accelerate convergence. Simultaneously, the cross-generational memory of the elite archive is maintained to prevent the loss of excellent solutions, and a quota mechanism ensures the number of first-frontier solutions.

[0071] Understandably, this embodiment generates weight values ​​for each decision variable to accurately measure its impact on the optimization objective, thus achieving targeted optimization. During iterative optimization, Gaussian perturbations are applied to key variables, the strength of which is proportional to the weight value and its range, decreasing with each iteration, effectively balancing global exploration and local search capabilities. This improvement makes path planning more flexible and efficient under complex constraints, further reducing travel time and energy consumption, significantly enhancing the operational efficiency and task execution capabilities of land-air robots in complex environments, and optimizing the overall performance of air-ground cooperative paths.

[0072] In some embodiments, during each iteration of optimization, based on a preset reference vector, the target variable that best approximates the Pareto front direction indicated by the reference vector is determined, forming a first candidate solution set including each target variable; based on the value of each variable in the current iteration of optimization, a normalized hyperplane is generated; each variable in the first candidate solution set is mapped to the normalized hyperplane for comparison, so as to select the variable for the next iteration of optimization.

[0073] In this embodiment, the solution can be implemented in two stages. In the first stage, the reference vector guides the selection, prioritizing the retention of solutions that approximate the frontier. In the second stage, a strict selection mechanism based on ideal hyperplane normalization is used to maintain solution set coverage and uniform distribution.

[0074] Understandably, this embodiment significantly improves the optimization efficiency and accuracy of path planning by introducing reference vectors and a normalized hyperplane mechanism. In each iteration, the target variable closest to the Pareto front direction is determined based on the reference vector, forming the first candidate solution set. These variables are then mapped to the normalized hyperplane for comparison and selection. This improvement effectively enhances the targeting and convergence speed of the optimization process, ensuring that key variables receive focused attention during optimization, while avoiding ineffective iterations. Ultimately, this technique further reduces travel time and energy consumption, significantly enhancing the operational efficiency and task execution capabilities of land-air robots in complex environments, and optimizing the overall performance of air-ground cooperative paths.

[0075] Please see Figure 4 This technical solution also provides a specific embodiment of an amphibious path planning method for land and air robots implemented through code, such as... Figure 4 As shown, in step S1, parameters such as the starting point, ending point, DEM (Digital Elevation Model), and road network are first input, and variables such as N (number of iterations), B (number of candidate solutions), α (control parameter), and w (weight) are set. Next, in step S2.4, path planning is performed to generate and store the expected ground-based planned path.

[0076] The process then enters a decision node, querying whether the destination has been reached. If not, it proceeds to the S3 path loop progression step. In this step, an aerial branch is first performed in S3.1 to determine the takeoff point, and then a VTOL (Vertical Takeoff and Landing) aerial trajectory is generated. Next, a ground branch is performed in S3.2 to calculate the path from the current node to the destination, update the current node as the destination, and return to the S3 loop.

[0077] If the path length Lh is determined to be less than a certain threshold in S3.2, then the expected ground planning path is generated in S3.3, the ground segment is saved, the current node is updated, and then the process returns to S3 for looping. If the current node is within the expected planning path during the path looping process in S3, then the path length is calculated and the angle is adjusted in S3.2.

[0078] Finally, in step S4, the amphibious path sequence and its time and energy consumption information are output, and the process ends. If the current node is not in the expected planned path during the path loop in step S3, the updated path in step S1 is re-executed. The entire flowchart illustrates an iterative process that continuously optimizes the path to find the optimal amphibious path.

[0079] Users only need to input the starting point, ending point, DEM, and road network under given parameters (N, θ_th, R, α, w). The program outputs the amphibious route (ground segment + air segment) along with its time and energy costs. The output road network environment map with the amphibious route is shown below. Figure 5 As shown.

[0080] This technical solution also provides embodiments for optimizing the solution based on the geographical environment of different regions.

[0081] With time and energy consumption as dual objectives: (1) Set optimization variables: α∈[1°,90°], R∈[40,250], θ_th∈[1°,90°], N∈[10,60], w∈[1,10] (the range can be adjusted according to engineering experience); (2)Ta MaOEA initializes the population, with each individual representing a set of parameters; (3) For each individual, call the above-mentioned amphibious path generation and calculate (time, energy consumption); (3) After two-stage selection and task-aware operator iterative update, the Pareto optimal parameter set and corresponding path set are output.

[0082] This embodiment can automatically adapt to different terrains (canyons / high altitudes / low altitudes) and output multiple executable solutions with different trade-offs for the task to choose from.

[0083] The results of energy consumption and travel time under different terrain conditions are as follows: Figure 6 ,7 As shown in Figures 8 and 9, the results of multi-objective amphibious planning in different altitude regions are presented. Figure 6 The corresponding terrain is the Wenchuan area (canyon). Figure 7 The corresponding terrain is the Tingri region (high altitude). Figure 8 The corresponding terrain is the Daling Mountain area (low altitude).

[0084] Please refer to Figure 9 , Figure 9 This is a schematic diagram of an amphibious path planning device for a land-air robot.

[0085] This embodiment also provides an amphibious path planning device for a land-air robot, comprising: The mapping module 201 is used to generate a road network environment map based on geographic data and road network data. The road network environment map includes multiple nodes, and each node is distributed according to preset rules to form a road network.

[0086] The setting module 202 is used to determine the current starting point and the target endpoint from multiple nodes in the road network.

[0087] The planning module 203 is used to generate a land path from the current starting point to the target endpoint along the node sequence arranged in a set order in the road network; take the target node other than the target endpoint in the land path as the take-off point and determine the landing point according to the preset search method to generate a flight path from the take-off point to the landing point; take the landing point as the current starting point and return to the step of generating a land path from the current starting point to the target endpoint along the node sequence arranged in a set order in the road network.

[0088] Output module 204 is used to output an amphibious path that includes all land paths and all flight paths when the current starting point is located at the target endpoint.

[0089] It will be understood by those skilled in the art that all or some of the steps and apparatuses in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. As is known to those skilled in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0090] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0091] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement any of the above-mentioned amphibious path planning methods for land and air robots.

[0092] It is understood that the content of the above method embodiments is applicable to the embodiments of this electronic device. The specific functions implemented by the embodiments of this electronic device are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0093] This application also provides a computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to implement the amphibious path planning method for land and air robots as described in any of the above specific embodiments.

[0094] This application also discloses a computer program product, including a computer program or computer instructions, which are stored in a computer-readable storage medium. The processor of the computer device reads the computer program or computer instructions from the computer-readable storage medium and executes the computer program or computer instructions, causing the computer device to perform the amphibious path planning method for land and air robots as described in any of the preceding embodiments.

[0095] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0096] The terms “first,” “second,” “third,” “fourth,” etc. (if applicable) used in this application's specification are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices. It should be understood that in this application, “at least one” means one or more, and “more than one” means two or more.

[0097] In the several embodiments provided in this application, it should be understood that the disclosed apparatus, devices, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0098] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0099] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0100] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0101] Although the description of this application has been quite detailed and particularly focused on several of the described embodiments, it is not intended to limit itself to any of these details or embodiments or any particular embodiment. Rather, it should be considered as effectively covering the intended scope of this application by referring to the appended claims and taking into account the prior art, which provides for a broad possible interpretation of these claims. Furthermore, the foregoing description of this application with respect to embodiments foreseeable by the inventors is intended to provide a useful description, and non-substantial modifications to this application that have not yet been foreseen may still represent equivalent modifications.

Claims

1. An amphibious path planning method for a land-air robot, characterized in that, The method includes: Based on geographic data and road network data, a road network environment map is generated, wherein the road network environment map includes multiple nodes, and each node is distributed according to preset rules to form a road network; Determine the current starting point and the target destination from multiple nodes of the road network; A land route from the current starting point to the target destination is generated along a sequence of nodes arranged in a predetermined order in the road network; Using target nodes other than the target endpoint in the land route as take-off points, and determining landing points according to a preset search method, a flight path from the take-off point to the landing point is generated; Using the landing point as the current starting point, return to the step of generating a land route from the current starting point to the target endpoint by following the node sequence arranged in a set order in the road network; When the current starting point is located at the target endpoint, the output includes an amphibious path that includes all the land paths and all the flight paths.

2. The method according to claim 1, characterized in that, Before determining the takeoff point, the method further includes: Calculate the terrain detour degree for each location on the land route, wherein the terrain detour degree is used to quantify the degree of detour of the path from the location along the road network to the next node; When the terrain detour exceeds a preset threshold, a target node is determined in the land travel path.

3. The method according to claim 2, characterized in that, The specific methods for determining the target node in the land travel path include: The direction from the current starting point to the target endpoint is used as the reference direction; The land route is divided into multiple segments according to a preset step length, wherein each preset step length includes at least one node. After determining the offset of each road segment, the cumulative heading offset is obtained by accumulating the offsets, where the offset is the offset of the heading angle of the road segment relative to the reference direction; When the cumulative heading offset exceeds the preset offset threshold for the first time, the node in the current step size is taken as the target node.

4. The method according to claim 2, characterized in that, The specific methods for determining the landing point based on the preset search method include: Using the takeoff point as a reference point, a sector-shaped area is established, wherein the centerline of the sector-shaped area points towards the target endpoint, and the sector-shaped area has a preset sector angle and search radius; If no node is present within the sector area, the search radius is increased, the angle range of the sector angle is adjusted, or the generation of the flight path is stopped. If at least one of the nodes exists within the sector area, then each of the nodes located within the sector area is considered a potential landing point; Based on the flight cost from the starting point to each of the potential landing points, a target landing point is selected from a plurality of potential landing points.

5. The method according to claim 4, characterized in that, The method further includes parameter optimization operations on key parameters, wherein the key parameters are the preset threshold, the search radius, or the sector angle, and the implementation methods of the parameter optimization operations include: Combine the key parameters into a decision variable set; A multi-objective optimization model is constructed with the goal of reducing travel time and energy consumption. The decision variable set is co-optimized using the multi-objective optimization model to obtain a Pareto optimal solution set, where each Pareto solution contains a set of optimized decision variable values.

6. The method according to claim 5, characterized in that, The specific methods for using the multi-objective optimization model to perform collaborative optimization of the decision variable set include: Generate a weight value for each variable in the decision variable group, wherein the weight value is used to characterize the degree of influence of the variable on the optimization objective; The decision variable group is iteratively optimized multiple times, and after each iteration, the optimized variables with weight values ​​higher than the preset level value are targeted for optimization. When performing a mutation operation on the target variable during each iteration of optimization, a Gaussian perturbation is applied to the target variable, wherein the strength of the Gaussian perturbation is proportional to the weight value corresponding to the target variable and the range of parameter values ​​of the target variable, and the strength of the Gaussian perturbation decreases as the number of iterations increases.

7. The method according to claim 6, characterized in that, In each iteration of optimization, the method further includes: Based on a preset reference vector, the target variable that best approximates the Pareto front direction indicated by the reference vector is determined, forming a first candidate solution set including each target variable; Generate a normalized hyperplane based on the value of each variable in the current iteration of optimization; Each variable in the first candidate solution set is mapped to the normalized hyperplane for comparison, in order to select the variables for the next iteration of optimization.

8. An amphibious path planning device for a land-air robot, characterized in that, The device includes: The mapping module is used to generate a road network environment map based on geographic data and road network data. The road network environment map includes multiple nodes, and each node is distributed according to preset rules to form a road network. The setting module is used to determine the current starting point and the target endpoint from multiple nodes of the road network; The planning module is used to generate a land route from the current starting point to the target endpoint along a sequence of nodes arranged in a preset order in the road network; using target nodes other than the target endpoint in the land route as take-off points and determining landing points according to a preset search method, a flight path from the take-off point to the landing point is generated; using the landing point as the current starting point, the module returns to the step of generating a land route from the current starting point to the target endpoint along a sequence of nodes arranged in a preset order in the road network. The output module is used to output an amphibious path that includes all the land paths and all the flight paths when the current starting point is located at the target endpoint.

9. An electronic device, characterized in that, 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 amphibious path planning method for the land and air robot according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the amphibious path planning method for the land and air robot as described in any one of claims 1 to 7.

Images