Low-altitude space path planning method and system

Abstract

The invention provides a low-altitude space path planning method and system, and relates to the technical field of low-altitude space management and path planning, and the method comprises the steps: S1, dividing a target airspace into a plurality of levels of three-dimensional voxels, and generating a unified-level three-dimensional code for each voxel; s2, voxelizing a task starting point, a task ending point and constraint elements based on unified hierarchical three-dimensional coding to generate a confinement three-dimensional corridor so as to limit a subsequent search space; s3, in the confinement three-dimensional corridor, hierarchical path search from the coarse hierarchy to the target hierarchy is executed, and a discrete path is output; s4, when the dynamic change of the environment is detected, performing increment re-planning on the basis of the discrete path, and updating the affected local path; and S5, smooth processing and safety verification are carried out on the discrete path or the updated path, and a final continuous flight path is generated. According to the invention, the real-time performance, the safety and the verifiability of the low-altitude flight task path can be realized.

Description

Technical Field

[0001] This invention relates to the field of low-altitude space management and route planning technology, specifically to a method and system for planning low-altitude space routes. Background Technology

[0002] Currently, low-altitude flight missions such as urban drone logistics and emergency inspections place extremely high demands on the real-time performance, safety, and verifiability of path planning. Existing technical solutions, such as continuous 3D geometric planning, 2D path overlay height, 2D geographic grids (e.g., S2, H3), and traditional octet voxel methods, all have significant shortcomings: they are either computationally complex and difficult to implement in real time, cannot guarantee 3D obstacle avoidance, lack unified height processing capabilities, or have low efficiency in hierarchical jumps. Especially in dynamically changing environments, existing solutions usually require global replanning, which is computationally expensive and prone to causing severe path jitter.

[0003] Therefore, there is an urgent need for a low-altitude path planning method that can uniformly represent three-dimensional space, support efficient cross-level operations, and enable incremental real-time replanning.

[0004] Explanation of related terms:

[0005] Voxel: A three-dimensional mesh cell at level L, whose spatial coverage in the longitude, latitude, and altitude directions is determined by coding.

[0006] Level (L): An integer representing the level of 3D mesh refinement. The higher the level, the smaller the voxel space scale.

[0007] OctID: A unique identifier for a voxel at each level, using a “bit-interleaved hierarchical three-dimensional encoding” (preferably three bits per level: 2 bits for the plane + 1 bit for the height).

[0008] Hmax: The upper limit of the uniform height domain (e.g., 65,536,000m, used for binary search to determine the height position).

[0009] Sign: Altitude sign bit, 0 indicates above-ground altitude, 1 indicates underground depth (positive values ​​are used in the binary search).

[0010] Corridor: A confined set of three-dimensional voxels for the search, consisting of a two-dimensional overlay (such as a QuadKey / GeoSOT planar mesh) and the Cartesian product of the height band, with a headroom buffer applied.

[0011] Hard constraints: Rules that cannot be violated, such as no-fly zones, minimum clearance, maximum climb rate, maximum tilt angle, and legal upper and lower altitude limits.

[0012] Cost terms: Weighted metrics used for optimization, such as horizontal travel, vertical energy consumption, wind / turbulence penalty, communication dead zone penalty, noise sensitivity weight, obstacle proximity potential field, etc.

[0013] Neighborhood (6 / 26 neighbors): The set of voxels that are adjacent to the current voxel at a fixed level L in terms of face / edge / corner.

[0014] 3D Traversal Verification: The voxel sequence traversed along the path segment is checked voxel by voxel to verify that no obstacle / buffer set has been entered. Summary of the Invention

[0015] To address the shortcomings of existing technologies, this invention provides a method and system for planning low-altitude spatial paths.

[0016] According to the present invention, a method and system for planning low-altitude space paths are provided, the scheme of which is as follows:

[0017] Firstly, a method for planning low-altitude spatial paths is provided, the method comprising:

[0018] Unified hierarchical 3D coding step S1: Divide the target airspace into several levels of 3D voxels, generate a unified hierarchical 3D code for each voxel, the code supports rapid positioning and jumping of parent voxels, child voxels and ancestor voxels, and can realize unified mapping between above-ground and underground spaces.

[0019] Step S2 of constructing a confined 3D corridor: Based on the hierarchical 3D encoding, the task start point, end point and related constraint elements are voxelized. Combined with the 2D plane coverage and allowable height range of the task path, a confined 3D corridor containing a safety buffer is generated to limit the subsequent search space.

[0020] Hierarchical path search step S3: Within the confined 3D corridor, first perform a skeleton-level search at the coarse level to obtain a skeleton path, and then use the skeleton path as a guide to perform a refined search on the region between adjacent voxel pairs within the target level to generate a discrete path at the target level.

[0021] Incremental replanning step S4: When dynamic changes in the environment are detected, incremental replanning is performed based on the discrete paths to update the affected local paths;

[0022] Path smoothing and safety verification step S5: Smooth the discrete path or the updated path and perform safety verification to generate a continuous and flyable final path.

[0023] Preferably, in the unified hierarchical 3D encoding step S1, the generation method of the hierarchical 3D encoding includes:

[0024] Each level of voxel uses a three-segment encoding structure consisting of planar position bits and height bits;

[0025] In the vertical direction, a layer-by-layer binary method based on sign bit and absolute height is adopted to map the above-ground and underground areas to the same height domain, thereby forming the unified hierarchical three-dimensional code.

[0026] When storing the unified hierarchical 3D code, it is divided into two storage areas, a high-order segment and a low-order segment, according to the bit order. The storage method supports sequential read and write operations that cross the boundaries of conventional data length, and voxel positioning and jumping can be achieved without unpacking all the unified hierarchical 3D codes.

[0027] Preferably, the height mapping of the unified hierarchical three-dimensional encoding achieves continuous encoding of above-ground and underground by taking the absolute values ​​of the above-ground height and underground depth in a unified height master domain and distinguishing them by the sign bit.

[0028] Preferably, in the restricted three-dimensional corridor construction step S2, the constraint elements include the flight allowable range, minimum airspace, regulatory height limit, and obstacle zone;

[0029] The generation of the safety buffer is based on the horizontal and vertical safety radii required by the task. The horizontal and vertical safety radii are converted according to the voxel size at different levels to ensure that the physical scale of the safety buffer is consistent at each resolution level.

[0030] The voxelization process is based on the unified hierarchical 3D encoding, which converts the task start point, end point, and constraint elements into corresponding voxel sets.

[0031] Preferably, in the hierarchical path search step S3, the skeleton-level search result includes path nodes and their cumulative cost, heuristic cost, and parent-child node pointer information. This information is used as the initial state input when refining the search to accelerate local search convergence.

[0032] The hierarchical search process in this step satisfies the following conditions:

[0033] Voxel hard constraint filtering: Remove non-compliant voxels (i.e. voxels that do not meet flight safety or regulatory requirements). Non-compliant voxels include voxels in no-fly zones, voxels below the minimum clearance range, voxels exceeding the legal altitude limit, and voxels that violate the maximum rate of climb or maximum climb angle conditions.

[0034] Cost term comprehensive calculation: Set a cost term for each candidate voxel node. The cost term includes horizontal distance cost, height change cost, wind field penalty, communication blind zone penalty, communication noise penalty, and safety potential energy cost based on the potential field of neighboring voxels.

[0035] Comprehensive evaluation mechanism: The cumulative cost of voxels is used as the path evaluation index. The candidate paths are sorted by a heuristic function. Under the premise of satisfying safety constraints, the feasible path with the minimum total cost is obtained. The feasible path is the high-precision discrete path at the target level.

[0036] Preferably, in the incremental replanning step S4, the dynamic environmental changes include: dynamic obstacles, wind field changes, temporary airspace adjustments, and changes in communication status; the triggering conditions for incremental replanning include: the appearance or disappearance of dynamic obstacles, sudden changes in wind speed or wind direction, temporary adjustments to the mission airspace, changes in remaining battery power or endurance, and communication signal attenuation.

[0037] This step specifically includes:

[0038] Time-varying voxel modeling: Dynamic entities are modeled as time-varying voxel sets in time slices, and the hard constraint parameters and cost terms of the affected voxels are updated synchronously within the confined 3D corridor to achieve local response to the dynamic environment.

[0039] Incremental replanning: When a change in voxel attributes is detected in the time-varying voxel set, incremental search and cost re-evaluation are performed only on the local regions adjacent to the batch of voxels, and the affected path segments are repaired while maintaining global path connectivity, so as to quickly obtain the updated optimal flight path.

[0040] Preferably, in the path smoothing and safety verification step S5, the path smoothing process includes compression of polyline corner points, merging of path segments, and curve smoothing based on constraints of maximum climb rate, maximum tilt angle, and minimum turning radius.

[0041] This step specifically includes:

[0042] Path smoothing: Based on the discrete path, curve fitting, gradient constraints, or spline interpolation are used to generate a continuous flight curve that conforms to the kinematic characteristics of the UAV.

[0043] Execution verification: The dynamic constraints of velocity, acceleration, turning radius and attitude changes of the continuous flight curve are tested to verify whether they meet the real-time executability requirements of the flight control system;

[0044] Safety verification: Within the confined three-dimensional corridor, the clearance distance, communication connectivity, and obstacle avoidance conditions of the continuous paths corresponding to the continuous flight curves are verified. Specifically, a three-dimensional traversal verification is performed on the continuous paths corresponding to the continuous flight curves. The three-dimensional traversal verification includes enumerating the voxel sequence it traverses along the path segment in sequence and detecting whether it enters an obstacle voxel or obstacle buffer set. If any indicator does not meet the preset threshold, a backtracking correction or local replanning is triggered.

[0045] Secondly, a low-altitude space path planning system is provided, the system comprising:

[0046] 3D encoding module: used to discretize 3D space into voxels and generate a unified hierarchical 3D code, which supports rapid voxel positioning and jumping and unified mapping between above-ground and underground spaces;

[0047] Corridor generation module: Based on the unified hierarchical 3D encoding, it performs voxelization processing on the task start point, end point and constraint elements to generate a confined 3D corridor containing a safety buffer.

[0048] Layered search module: used to sequentially perform coarse-level skeleton-level search and target-level refinement search within the confined three-dimensional corridor, and output a high-precision discrete path;

[0049] Incremental replanning module: used to model dynamic factors as a time-varying set of voxels. When the voxel attributes in the time-varying set are updated, the local segments of the path are re-evaluated and replaced to generate the updated optimal path.

[0050] Path smoothing verification module: used to smooth and verify the discrete path or the updated path, and generate a continuous and flyable final path.

[0051] Preferably, the system further includes: a constraint and cost module, which includes a hard constraint removal unit, a soft constraint cost unit, and a dynamic obstacle management unit;

[0052] The hard constraint rejection unit is used to reject non-compliant voxels, including voxels in no-fly zones and voxels below the minimum airspace clearance range.

[0053] The soft-constraint cost unit is used to calculate the cost term of the candidate voxel node, and the cost term includes horizontal distance cost and height change cost.

[0054] The dynamic obstacle management unit is used to model dynamic factors as a time-varying set of elements.

[0055] Preferably, the soft constraint cost unit is further used to calculate the wind field penalty term based on the wind field model, the communication blind zone penalty term and the communication noise penalty term based on the communication coverage model, and the safety potential energy cost based on the neighboring voxel potential field model, and to weight and superimpose the above cost terms into the comprehensive cost of the candidate voxel node.

[0056] Thirdly, a computer-readable storage medium storing a computer program is provided, which, when executed by a processor, implements the steps of the low-altitude space path planning method.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] 1. This invention achieves continuous representation of above-ground and underground space through unified hierarchical three-dimensional coding, solving the problem of discontinuous coding near the ground surface near zero height in traditional schemes;

[0059] 2. This invention achieves real-time access across levels by using high-low segmented storage and fast jump, thereby improving the efficiency of searching and neighborhood expansion;

[0060] 3. This invention reduces the search space from the entire domain to a finite volume by using a corridor generation method that combines two-dimensional coverage with height bands and a safety buffer, thus significantly reducing computational complexity.

[0061] 4. This invention improves search speed while ensuring path accuracy through a skeleton-guided hierarchical search and hot-start mechanism;

[0062] 5. This invention maintains path connectivity and reduces replanning jitter in dynamic environments through a local incremental replanning mechanism; the replanning adopts a time-varying element and incremental search mechanism, which can respond to environmental changes in milliseconds and achieve local path adaptation.

[0063] 6. This invention ensures the security and compliance of the output path through a verifiable mechanism that combines smoothing and voxel verification;

[0064] 7. The unified hierarchical three-dimensional coding proposed in this invention simultaneously represents planar position and height information in the same coding space. It retains the advantages of hierarchical planar grids in terms of large-scale coverage and fast indexing, and directly generates continuous three-dimensional voxel partitions at the coding level through the layer-by-layer bisection of height bits. This fundamentally eliminates the fragmented mode of "two-dimensional grid + external height" and realizes true three-dimensional integrated path planning.

[0065] Other beneficial effects of the present invention will be explained in detail through the introduction of specific technical features and technical solutions in specific embodiments. Those skilled in the art should be able to understand the beneficial technical effects brought about by these technical features and technical solutions through the introduction of these technical features and technical solutions. Attached Figure Description

[0066] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0067] Figure 1 This is a flowchart illustrating the low-altitude space path planning method. Detailed Implementation

[0068] The present invention will be described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several changes and improvements can still be made. These all fall within the protection scope of the present invention.

[0069] An embodiment of the present invention provides a method for planning a low-altitude space path. Referring to Figure 1 as shown, the method specifically includes:

[0070] S1. Unified hierarchical three-dimensional coding step:

[0071] This step establishes an efficient and unified three-dimensional space data expression model for the target airspace.

[0072] First, the target airspace is discretized into three-dimensional voxels at multiple levels (L). The level L is an integer. The larger the level, the smaller the spatial scale of the voxel. For example, level L = 0 can represent the entire earth space, while level L = 26 can achieve meter-level voxel accuracy.

[0073] Then, a unique code (OctID) at this level is generated for each voxel. This code is concatenated from high to low by L "three-bit segments". Each "three-bit segment" is formed by combining the planar position bits (XYdigit) and height bits (Zbit) at this level. Specifically:

[0074] Planar position bits: By performing a quadtree division on the longitude and latitude plane at this level, the quadrant values (0, 1, 2, 3) are obtained.

[0075] Height bits: Obtained by the逐层二分法 (layer-by-layer binary division) in the height direction. The specific process is as follows: In a unified height main domain [0, Hmax], take the absolute value of the above-ground height, take the positive value of the underground depth, and use a sign bit (Sign) to distinguish. Let the current binary division interval be [lo, hi], and take the midpoint mid = (lo + hi) / 2. If the absolute height absH of the current point < mid, then the height bit Zbit of this layer = 1, and let hi = mid; otherwise, Zbit = 0, and let lo = mid. After iterating L times, a Zbit sequence with a length of L is obtained, thus achieving continuous and unified coding of the above-ground and underground spaces.

[0076] Finally, in order to efficiently store and process the possibly up-to 96-bit (corresponding to 32 levels) code, the generated unique code is stored in two storage areas, the high segment (Hi) and the low segment (Lo), in segments according to the bit order. This storage structure supports fast hierarchical jump operations:

[0077] Obtaining the parent / ancestor voxel: This is achieved directly by truncating the high-order prefix of the encoding (i.e., removing the last k three-digit segments).

[0078] Enumerating sub-voxels: This is achieved directly by appending a three-dimensional field (val∈{0...7}) to the end of the encoding. These operations are all constant time O(1), providing efficient underlying support for subsequent steps.

[0079] S2. Steps for constructing a confined 3D corridor:

[0080] This step, based on the unified hierarchical 3D encoding from the previous step, transforms the task elements of the physical world into a discrete, constrained search space.

[0081] First, voxel representation is performed. The task start and end points, as well as various constraints (such as buildings, no-fly zones, height-restricted areas, terrain, etc.), are all converted into voxel sets at the corresponding levels. This includes:

[0082] Two-dimensional coverage: Based on a unified hierarchical three-dimensional encoding, a set of candidate two-dimensional cells is obtained by covering the line connecting the task start point and the task end point and its extended range with a planar grid.

[0083] Height band padding: The allowed height range [Hmin, Hmax_task] of the task is converted into a height index band at the coding level, and then Cartesian product is performed with the above two-dimensional cell set to obtain the initial three-dimensional voxel set.

[0084] Corridor width and clearance buffer: Based on the minimum clearance (d_clear) and safety boundary required by the mission, set the horizontal safety radius (R_xy) and the vertical safety radius (R_z). Convert these two physical radii to voxel step radii at the corresponding level L, and perform neighborhood expansion (buffer expansion) on the initial voxel set to form the final confined 3D corridor C. Simultaneously, obstacles and no-fly zones also undergo the same buffer expansion to form a no-fly buffer set.

[0085] Secondly, generate neighborhoods and buffers. Under a fixed level L, define a set of first-order neighborhoods (such as 6-neighborhoods or 26-neighborhoods) that are adjacent to the current voxel in the face, edge, and corner directions, which are used for connectivity determination during subsequent path search.

[0086] S3, Hierarchical path search steps:

[0087] This step involves efficient hierarchical path search within the confined 3D corridor C.

[0088] First, a coarse-level skeleton search is performed (at level L0). Inputs include: the voxel set of the confined corridor C at level L0, the voxel identifiers of the task's start and end points at L0, and voxel-level hard constraints and cost terms. A search algorithm is used to search the corridor subgraph at level L0. During the search, hard constraints (such as no-fly zones, insufficient airspace, exceeding legal altitude, exceeding maximum climb rate / tilt) are directly used for filtering, preventing the search from entering non-compliant voxels. Cost terms (such as horizontal distance, altitude change, wind field penalty, communication blind spot penalty, noise penalty, obstacle proximity potential field) are used to calculate the movement cost G(n) and heuristic function H(n) to guide the search towards low-cost, feasible directions. The final output is a skeleton path (a sequence of voxels at level L0) and cost records for incremental replanning (such as the cumulative cost G, heuristic cost H, and parent node pointer for each node).

[0089] Then, a local refinement search is performed (at the target level L). Guided by the coarse-layer skeleton path, for each pair of adjacent voxels on the skeleton path, the fast jump capability of step S1 is used to determine their candidate set of sub-voxels at the target level L. The search algorithm is run again within the intersection of this candidate set and corridor C. This process can inherit the overhead record of the coarse-layer search as a warm start, thereby accelerating convergence. Finally, all locally refined path segments are concatenated to generate a high-precision discrete path P_L at the target level L.

[0090] S4. Incremental replanning steps:

[0091] This step is used to update the path quickly and smoothly when the environment changes dynamically.

[0092] The triggering conditions include: dynamic obstacle updates, sudden changes in wind field, temporary airspace adjustments, changes in remaining range / battery power, and significant attenuation of communication links.

[0093] The specific process includes:

[0094] Time-varying voxel modeling: Dynamic entities (such as moving drones, temporary no-fly zones, and changing wind fields) are modeled as sets of voxels (x, y, z, t) that are updated over time slice (t). Within a confined 3D corridor, the hard constraints and cost terms of the spatial regions affected by these time-varying voxels are updated in real time.

[0095] Incremental Repair: Based on the existing discrete path P_L and its cost records, when a batch of voxel attributes changes are detected, only the local regions adjacent to these changed voxels are re-estimated. Using an incremental search approach similar to LPA or D*Lite, the G / H values ​​of nodes are locally relaxed again. If the original path is blocked, only the affected road segments are backtracked and re-searched; unchanged path segments remain unchanged. This mechanism significantly reduces replanning time and effectively reduces path jitter.

[0096] S5. Path smoothing and security verification steps:

[0097] This step transforms discrete voxel paths into smooth, safe trajectories that can be directly executed by the UAV.

[0098] Path smoothing:

[0099] Corner compression and polyline merging: The discrete path P_L* is scanned, and redundant corner points that are collinear or approximately collinear are removed within the allowable error threshold and clearance constraints. Furthermore, attempts are made to establish visible direct connections between nodes. If the direct connection segment passes the safety check, a single segment replaces the multi-segment polyline, thus simplifying the path.

[0100] Curve smoothing: The compressed polyline is fitted with a spline curve to generate a smooth trajectory that is continuous in terms of position, velocity, and acceleration.

[0101] Time parameterization: Distribute velocity profiles along the curve arc length to ensure that the trajectory meets the dynamic constraints of the UAV, such as maximum climb rate, maximum tilt angle, and minimum turning radius.

[0102] Execution verification and security review:

[0103] 3D Traversal Verification: For each smoothed path segment, enumerate all voxels it traverses sequentially according to its spatial location. If any voxel traversed belongs to an obstacle voxel or a no-fly buffer set, or does not meet the minimum clearance requirements, then the path segment is deemed unsafe.

[0104] Dynamic constraint detection: Verify whether the time-parameterized trajectory strictly satisfies all preset dynamic envelopes.

[0105] Indicator Output and Closed Loop: Calculate and output auditable safety indicators, including minimum airspace clearance, distance to the nearest no-fly zone, and maximum climb / descent rate satisfaction. If all checks pass, output the final flyable trajectory curve; otherwise, feed back the problematic section to the preceding step S3 or step S4 to trigger local replanning, forming a safety closed loop.

[0106] Implementation Case: Urban Low-Altitude Logistics Scenarios;

[0107] Suppose we are carrying out a drone delivery mission in a certain urban area.

[0108] Region and level: Target level L* = 24 (approximately meter-level accuracy), coarse level L0 = 20.

[0109] Constraints: Flight altitude zone is [60m, 150m], minimum clearance is 5m, maximum climb rate is 6m / s.

[0110] Specific implementation:

[0111] S1: Using geographic information system data, construct a three-dimensional voxel code for the region from L0 to L*.

[0112] S2: The starting point is warehouse A, and the ending point is distribution point B. Extending 50 meters outward from the line connecting them forms a two-dimensional coverage, which is combined with the [60,150] meter height zone and a 5-meter clearance buffer is provided for external building elements to generate a confined corridor.

[0113] S3: Quickly search for a skeleton path that bypasses the core urban area at the L0 level, and then refine it at the L* level to obtain a discrete path that precisely avoids all buildings.

[0114] S4: During flight, a temporary no-fly zone (such as an active airspace) suddenly appears. The system models it as a time-varying element, immediately triggers incremental replanning, and generates a new path that bypasses the temporary no-fly zone within seconds, while most of the original path remains unchanged.

[0115] S5: Perform spline smoothing on the final path, verify that the total clearance is greater than 5 meters and the climb rate meets the requirements, and then output the result to the UAV for execution.

[0116] This invention also provides a low-altitude space path planning system, which can be implemented by executing the process steps of the low-altitude space path planning method. That is, those skilled in the art can understand the low-altitude space path planning method as a preferred embodiment of the low-altitude space path planning system. The system includes:

[0117] 3D encoding module: used to discretize 3D space into voxels and generate a unified hierarchical 3D code. The unified hierarchical 3D code supports fast voxel positioning and jumping, as well as unified mapping between above-ground and underground spaces.

[0118] Corridor generation module: Based on the unified hierarchical 3D encoding, it performs voxelization processing on the task start point, end point and constraint elements to generate a confined 3D corridor containing a safety buffer.

[0119] Layered search module: used to perform coarse-level skeleton search and target-level refinement search sequentially within a confined 3D corridor, and output a high-precision discrete path;

[0120] Incremental replanning module: This module models dynamic factors as a time-varying set of voxels. When the voxel attributes in the time-varying set are updated, the local segments of the path are re-evaluated and replaced to generate the updated optimal path.

[0121] Path smoothing verification module: Used to smooth and verify discrete or updated paths, generating a continuous and flyable final path.

[0122] Furthermore, the system also includes a constraint and cost module, which comprises a hard constraint elimination unit, a soft constraint cost unit, and a dynamic obstacle management unit;

[0123] Among them, the hard constraint elimination unit is used to eliminate non-compliant voxels, including voxels in no-fly zones and voxels below the minimum airspace clearance range;

[0124] The soft-constraint cost unit is used to calculate the cost terms for candidate voxel nodes. The cost terms include horizontal distance cost and height variation cost.

[0125] The dynamic obstacle management unit is used to model dynamic factors as a time-varying set of elements.

[0126] The present invention will now be described in more detail.

[0127] This invention provides a method for planning low-altitude space paths, referring to... Figure 1 As shown, each step is explained below:

[0128] 1. Unified hierarchical three-dimensional coding:

[0129] The target airspace is first divided into several levels of voxel units. Each voxel is determined by both planar coordinates and altitude index.

[0130] The height direction adopts a layer-by-layer binary division method, dividing the entire height range into multiple sub-intervals, and using a sign bit to distinguish between above-ground and underground, thereby achieving continuous coding near zero height.

[0131] During encoding, the height bits and plane bits of each layer are combined in a fixed order to form a three-bit segment. All three-bit segments from all layers are concatenated from high to low to form a unique code.

[0132] The encoding is stored in two bit-ordered memory segments: the higher-order segment stores high-level information, and the lower-order segment stores fine-level information. Parent or child voxels can be obtained in constant time by directly truncating or appending bits.

[0133] This structure ensures rapid positioning between different levels, providing support for subsequent corridor delimitation and hierarchical search.

[0134] 2. Construction of a confined 3D corridor:

[0135] After completing spatial coding, the system voxels all the starting point, ending point, and constraint elements of the task.

[0136] A two-dimensional covering set is formed based on the range of the line connecting the start and end points on the plane;

[0137] Map the allowed height range to the corresponding height index band;

[0138] The Cartesian product of the two forms a preliminary three-dimensional voxel set.

[0139] Subsequently, based on the required safety radius, the set is expanded in both the horizontal and vertical directions to obtain a three-dimensional corridor containing a safety buffer.

[0140] This corridor serves as the bounded space for subsequent path search, and its generation relies on the fast parent-child voxel switching capability provided by the encoding structure.

[0141] 3. Hierarchical path search:

[0142] The search process is divided into two levels.

[0143] At a coarser level, the system only considers the skeleton structure of the corridor and searches for an approximate skeleton path based on the accessibility and cost of the voxels.

[0144] Subsequently, at the target level, guided by the skeleton path, a refined search is performed on the space between adjacent voxel pairs of the skeleton path.

[0145] During refinement, the system uses the cost and heuristic information from the skeleton stage as initial values ​​for rapid expansion to obtain a high-precision discrete path.

[0146] The entire search process is conducted under corridor constraints, which significantly reduces invalid expansion.

[0147] 4. Incremental replanning:

[0148] The process comprises two sub-processes: time-varying voxel modeling and incremental replanning. During flight, if changes occur in the external environment, such as obstacle movement, sudden wind changes, or communication signal variations, the system updates the changed region with a new voxel set and re-estimates the local path regions adjacent to this set. Specifically, time-varying voxel modeling synchronously updates the constraints and cost parameters of the affected region on a time-slice basis, enabling rapid response to environmental changes.

[0149] This local update method avoids global recalculation, greatly improves response speed, and ensures smooth path transition, reducing flight jitter.

[0150] 5. Path smoothing and security verification:

[0151] It includes three sub-steps: path smoothing, executability verification, and security review.

[0152] Discrete paths are smoothed to generate a continuous trajectory that can fly.

[0153] During the smoothing process, redundant corner points in the path are first compressed and small-angle polyline segments are merged. Then, the polylines are transformed into smooth curves according to the constraints of aircraft dynamics.

[0154] The system performs voxel traversal verification before and after smoothing, that is, it checks whether the path traverses obstacle voxels or their buffers segment by segment.

[0155] If the path fully meets the safety requirements, the system outputs the final trajectory and generates auditable safety metrics, including minimum airspace clearance, closest distance to no-fly zones, and climb / descent rate compliance. If any safety metric fails to meet the threshold, the system will trigger a backtracking correction or replanning to form a closed-loop verification mechanism.

[0156] This invention provides a method and system for planning low-altitude spatial paths. By using a unified hierarchical three-dimensional encoding, it solves the problem of continuous representation and efficient indexing of above-ground and underground spaces. Through confined corridors and hierarchical search, it reduces computational complexity from global to local, enabling real-time planning. Through an incremental replanning mechanism, it ensures rapid response and path smoothness in dynamic environments. Through rigorous path smoothing and voxel-level safety verification, it ensures the aircraft's executability and safety verifiability of the final output path, providing reliable technical support for low-altitude intelligent flight.

[0157] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0158] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method of planning a low airspace path, characterized by, The method comprises the following steps: A uniform hierarchical three-dimensional coding step S1: dividing a target space into several hierarchical three-dimensional voxels, and generating a uniform hierarchical three-dimensional code for each voxel, wherein the code supports quick positioning and jumping of parent voxels, child voxels and ancestor voxels, and can realize uniform mapping of above-ground and underground spaces; A limited three-dimensional corridor construction step S2: based on the uniform hierarchical three-dimensional code, voxelizing a task starting point, a task ending point and related constraint elements, and combining a two-dimensional plane coverage range and an allowed height range of a task path to generate a limited three-dimensional corridor containing a safety buffer zone, so as to limit a subsequent search space; A hierarchical path search step S3: only in the limited three-dimensional corridor, performing a skeleton-level search at a coarse hierarchical level to obtain a skeleton path, and then performing a refinement search on a region between adjacent voxel pairs in a target hierarchical level to generate a discrete path in the target hierarchical level, with the skeleton path as a guide; An incremental re-planning step S4: when detecting dynamic changes in an environment, performing incremental re-planning based on the discrete path to update a local path affected; A path smoothing and safety verification step S5: performing smoothing processing and safety verification on the discrete path or the updated path to generate a continuous and flyable final path.

2. The low-space-path planning method according to claim 1, characterized in that, In the uniform hierarchical three-dimensional coding step S1, the generation mode of the hierarchical three-dimensional code comprises the following steps: Each voxel at each hierarchical level adopts a three-bit segment coding structure composed of a plane position bit and a height bit; In the height direction, a two-layer dichotomy based on a sign bit and an absolute height is adopted to uniformly map above-ground and underground spaces to a same height main domain, and then the uniform hierarchical three-dimensional code is formed; When storing the uniform hierarchical three-dimensional code, the uniform hierarchical three-dimensional code is divided into a high-bit segment and a low-bit segment in a bit sequence, and the storage mode supports sequential read-write operations across a conventional data length boundary.

3. The method of planning a low-space path according to claim 2, wherein, The height mapping of the uniform hierarchical three-dimensional code is realized by taking absolute values of above-ground heights and underground depths in a uniform height main domain and distinguishing them by a sign bit, so as to realize continuous coding of above-ground and underground spaces.

4. The low-space-path planning method according to claim 1, wherein In the limited three-dimensional corridor construction step S2, the constraint elements comprise a flight allowed range, a minimum clearance, a regulation height limit and an obstacle area; The safety buffer zone is generated according to a horizontal safety radius and a vertical safety radius required by a task, and the horizontal safety radius and the vertical safety radius are converted according to a voxel size at different hierarchical levels, so as to ensure that physical dimensions of the safety buffer zone are consistent at different resolution hierarchical levels; The voxelization processing is based on the uniform hierarchical three-dimensional code, and converts the task starting point, the task ending point and the constraint elements into corresponding voxel sets.

5. The low-space-path planning method according to claim 1, wherein, In the hierarchical path search step S3, the skeleton-level search result comprises path nodes, cumulative costs, heuristic costs and parent-child node pointer information, and the information is input as an initial state in the refinement search to accelerate local search convergence; In the hierarchical search process of this step, the following conditions are met: Voxel hard constraint filtering: removing non-compliant voxels, wherein the non-compliant voxels comprise forbidden flight area voxels, voxels lower than a minimum clearance range, voxels exceeding a legal height limit, and voxels violating a maximum climb rate or a maximum climb inclination condition; Cost item comprehensive calculation: set a cost item for each candidate voxel node, which includes horizontal distance cost, height change cost, wind field penalty item, communication blind area penalty item, communication noise penalty item, and safety potential energy cost based on adjacent voxel potential field; Comprehensive evaluation mechanism: take the cumulative cost of voxels as the path evaluation index, combine the heuristic function to sort the candidate paths, and obtain the feasible path with the minimum total cost under the premise of meeting the safety constraint, which is the high-precision discrete path of the target level.

6. The low-space-path planning method according to claim 1, wherein, In the incremental re-planning step S4, the environmental dynamic changes include: dynamic obstacles, wind field changes, temporary airspace adjustments, and communication condition changes; the triggering conditions of incremental re-planning include: the appearance or disappearance of dynamic obstacles, the mutation of wind speed or direction, the temporary adjustment of task airspace, the change of remaining power or endurance, and the attenuation of communication signal; The step specifically includes: Time-varying voxel modeling: model dynamic entities into a time-varying voxel set in time slices, and simultaneously update the hard constraint parameters and cost items of the affected voxels in the limited three-dimensional corridor to realize local response to dynamic environment; Incremental re-planning: when detecting changes in voxel attributes in the time-varying voxel set, only perform incremental search and cost re-estimation on the local area adjacent to the batch of voxels, and repair the affected path segment under the premise of maintaining global path connectivity to quickly obtain the updated optimal flight path.

7. The low-space-path planning method according to claim 1, wherein In the path smoothing and safety verification step S5, the path smoothing process includes compression of polyline corner points, merging of path segments, and curve smoothing according to the constraints of maximum climb rate, maximum tilt angle, and minimum turning radius; The step specifically includes: Path smoothing processing: based on the discrete path, generate a continuous flight curve that meets the kinematic characteristics of the UAV by using curve fitting, gradient constraint, or spline interpolation; Executable verification: perform dynamic constraint detection on the continuous flight curve in terms of speed, acceleration, turning radius, and attitude change to check whether it meets the real-time executable requirements of the flight control system; Safety review: review the clearance distance, communication connectivity, and obstacle avoidance conditions of the continuous path corresponding to the continuous flight curve in the limited three-dimensional corridor; Specifically, perform three-dimensional traversal review on the continuous path corresponding to the continuous flight curve, which includes enumerating the voxel sequence traversed along the path segment in sequence, and detecting whether it enters an obstacle voxel or an obstacle buffer set. If any index does not meet the preset threshold, trigger backtracking correction or local re-planning.

8. A low-altitude space path planning system based on the low-altitude space path planning method according to any one of claims 1-7, characterized in that, It includes: Three-dimensional encoding module: used for discretizing three-dimensional space into voxels and generating a unified hierarchical three-dimensional code, which supports fast voxel positioning and jumping and unified mapping of ground and underground space; Corridor generation module: used for voxelizing the task starting point, ending point, and constraint elements based on the unified hierarchical three-dimensional code to generate a limited three-dimensional corridor containing a safety buffer; Hierarchical search module: used for sequentially performing coarse skeleton level search and target level refinement search in the limited three-dimensional corridor to output a high-precision discrete path; An incremental re-planning module is configured to model dynamic factors as a set of time-varying voxels, and re-evaluate and replace a local segment of the path when the attribute of a voxel in the set of time-varying voxels is updated, to generate an updated optimal path. A path smoothing and validation module is configured to smooth and validate the discrete path or the updated path, to generate a continuous and flyable final path.

9. The low-space-path planning system of claim 8, wherein, The system further comprises a constraint and cost module, which includes a hard constraint removal unit, a soft constraint cost unit, and a dynamic obstacle management unit. The hard constraint removal unit is configured to remove non-compliant voxels, including no-fly zone voxels and voxels below a minimum clearance range. The soft constraint cost unit is configured to calculate candidate voxel node cost terms, including horizontal distance cost and height change cost. The dynamic obstacle management unit is configured to model dynamic factors as a set of time-varying voxels.

10. The low-space-path planning system of claim 9, wherein, The soft constraint cost unit is further configured to calculate a wind field penalty term based on a wind field model, a communication blind area penalty term and a communication noise penalty term based on a communication coverage model, and a safety potential energy cost based on a neighboring voxel potential field model, and to weight and superimpose the above cost terms as a comprehensive cost of a candidate voxel node.

11. A computer readable storage medium storing a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the low-altitude space path planning method of any one of claims 1 to 7.

Images