A heterogeneous unmanned aerial vehicle swarm self-healing path planning method and system based on capability space modeling and semantic topology maintenance

Abstract

This invention discloses a self-healing path planning method and system for heterogeneous UAV swarms based on capability space modeling and semantic topology preservation, belonging to the field of UAV swarm collaborative command and control. Its core lies in elevating the constraint object of path planning from the physical geometric position of the UAV to a capability space semantic topology layer independent of the physical space. The method includes: establishing a capability space model for each UAV, abstracting it as a node capable of radiating multiple virtual mission capabilities; constructing a mission capability semantic topology graph using the UAV as the physical carrier and capability coverage relationships as directed edges; defining topological constraints applied to the attributes of this semantic topology graph, rather than constraints directly applied to physical positions; when physical node damage leads to the destruction of the semantic topology structure, resolving capability gaps within the capability space, and performing capability pattern reconstruction and physical position collaborative optimization based on the potential capability list of candidate nodes, achieving self-healing of the semantic topology through role switching and route replanning. This invention achieves continuous semantic-level preservation of mission capabilities in space under conditions of variable physical nodes and switchable roles, significantly improving the mission resilience of the swarm in adversarial environments.

Description

Technical Field

[0001] This invention belongs to the field of command and control technology for collaborative combat of unmanned aerial vehicles (UAVs), and specifically relates to a method and system for realizing swarm path self-healing planning by constructing and maintaining mission capability semantic topology under combat damage conditions. Background Technology

[0002] Heterogeneous drone swarms consist of drones carrying different mission payloads. Traditional path planning methods primarily focus on geometric collision-free operation, optimal range, or maintenance of communication connectivity in physical space, directly imposing constraints on the physical positions of the drones. This approach treats drones as homogeneous physical point masses, failing to express and handle the spatial distribution and correlation of mission capabilities.

[0003] When a drone within a swarm performing a specific function is damaged in combat, even if physical connectivity remains, the swarm's mission capabilities may have suffered structural defects. For example, blind spots may appear in the ring-shaped reconnaissance coverage protecting critical assets, or the coordination link between electronic suppression cover and reconnaissance guidance may be broken. Existing methods, lacking independent modeling of mission capabilities, cannot perceive and repair such damage at the capability level; they can only perform simple physical node replacements, which is insufficient to meet the resilient operational requirements of complex heterogeneous capability coordination.

[0004] Therefore, there is an urgent need for a method that elevates the planning perspective from physical space to capability space, enabling semantic-level task capability topology self-healing of heterogeneous swarms under combat damage conditions. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for self-healing path planning of heterogeneous UAV swarms based on capability space modeling and semantic topology preservation.

[0006] The core idea is to establish a "task capability space" independent of physical geometry as the second-dimensional abstract space for planning. Within this space, drones are considered nodes capable of radiating multiple virtual capability regions, and the cooperative relationships of the swarm are expressed as a "task capability semantic topology graph." The goal of path planning is no longer to maintain physical formation, but to maintain the integrity of this semantic topology graph through physical movement and capability pattern reconstruction. When physical nodes are lost, the system searches for reconfigurable potential capabilities within the capability space, activates them, and guides them to the required locations, thereby achieving continuous self-healing of the task at the semantic layer.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A self-healing path planning method for heterogeneous UAV swarms based on capability space modeling and semantic topology preservation includes steps S1 to S7 and corresponding system modules.

[0008] Compared with the prior art, the present invention has the following beneficial effects: A fundamental improvement in planning dimensions: The planning object is elevated from its physical geometric location to the semantic topology of the capability space, which solves the essential defect of traditional methods in being unable to express and maintain the collaborative relationship between task capabilities.

[0009] Decoupling and self-healing of physics and capabilities: When a physical node fails, its task capabilities can be inherited and spatially transferred by other nodes with potential capabilities through pattern reconstruction, truly achieving resilient collaboration that is "dispersed in form but not in spirit".

[0010] Unified expression and collaborative constraints of heterogeneous capabilities: It can define and maintain complex logical combination constraints of various heterogeneous capabilities such as reconnaissance, interference, and relay within a unified topology graph framework, which greatly expands the complexity and robustness of the bee colony's autonomous collaboration. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating the method of the present invention.

[0012] Figure 2 This is a schematic diagram illustrating the construction principle of the task capability space model and semantic topology graph in this invention.

[0013] Figure 3 This is a schematic diagram of the initial semantic topology and physical path, which includes constraints on the combination of heterogeneous capabilities.

[0014] Figure 4 This is a schematic diagram of the semantic topological gap analysis and self-healing replanning results after battle damage. Detailed Implementation

[0015] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0016] Example 1: Escort mission scenario considering heterogeneous capability combination constraints This embodiment uses a critical asset UAV U_A carrying electronic reconnaissance equipment performing a close-range reconnaissance mission, requiring full escort protection from a swarm. The escort swarm includes: pure reconnaissance UAVs U_R1 and U_R2, electronic jamming UAV U_J1, communication relay UAV U_C1, and multiple switchable-mode reconnaissance and strike UAVs U_S1 and U_S2 (with potential reconnaissance / ground attack capabilities) and multi-functional electronic warfare UAV U_E1 (with potential electronic jamming / reconnaissance capabilities).

[0017] Step 1: Task Capability Space Modeling For each drone, construct a capability space model independent of physical geometry according to the following formula:

[0018]

[0019] The capability types include: reconnaissance (S), electronic jamming (J), and communications relay (R). Coverage determination functions for each type are also provided. The shape and extent of the virtual capability region radiating from the physical location pipi are defined. For example, the interference capability J is a forward sector.

[0020] .

[0021] The initial states and capabilities of each machine are listed in the table below (described in text form): Steps 2-3: Semantic Topology Graph Construction and Heterogeneous Combination Constraint Definition Construct a semantic topology graph for task capabilities. U_A, as a key asset, needs to accept directed semantic edges of type S and type J from other nodes. The complex task capability topology constraints defined in this embodiment are as follows: Heterogeneous capability combination constraint: Key asset U_A must simultaneously satisfy:

[0022]

[0023] This constraint requires that U_A be covered by at least two reconnaissance nodes from different directions (angle difference ≥ 60°), and simultaneously covered by at least one jamming node, with a communication relay (R-type) semantic path connection between the node providing jamming coverage and at least one node providing reconnaissance coverage. This logical constraint of combining multiple edge types is something that no existing physical formation control method can define.

[0024] like Figure 3 As shown, the initial plan satisfies this constraint: U_R1 and U_R2 provide two-way reconnaissance coverage; U_J1 provides forward interference coverage; U_C1 acts as a relay, connecting U_J1 and U_R1 on the semantic graph.

[0025] Steps 4-5: Initial Planning and Battle Damage Monitoring Plan the physical paths of each aircraft to accommodate the aforementioned semantic topology. During flight, electronic jamming aircraft U_J1 is destroyed. The semantic topology immediately changes: jamming capability nodes are lost, J-type semantic edges disappear, leading to a violation of heterogeneous capability combination constraints.

[0026] Step 6: Semantic topology self-healing solution within the capability space The system triggered a corrective replanning. Capability gap analysis: One Class J capability node is missing, and this node needs to maintain relay connectivity with the reconnaissance node. The system queried the potential capability list (LatentCap) of the remaining aircraft and found that the multi-functional electronic warfare UAV U_E1 has potential jamming capabilities. There is an error; it should be corrected to {S,J}, as set above. Its potential ability is S, therefore it cannot be directly supplemented with J. (The setting needs to be corrected here).

[0027] The settings are revised to reflect more complex reconstructions: Let U_E1 be... U_S1 is At this point, the capability gap is "Type J capability". U_E1 possesses potential capability J and has been selected as a candidate aircraft. U_S1 possesses potential capability S and can be selected as a reconnaissance backup candidate.

[0028] Joint Optimization: The solver discovers a two-stage cascaded reconstruction: 1) Instructing U_E1 to switch from reconnaissance mode (ActiveS) to jamming mode (ActiveJ), flying to the replacement site of the original U_J1 to restore J-type coverage; 2) Simultaneously instructing U_S1 to switch from attack mode to reconnaissance mode, flying to fill the reconnaissance coverage gap left by U_E1. The objective function JJ weighs the performance loss of U_S1 abandoning the attack mission against the range cost of both aircraft, and incorporates a topology deviation penalty to maintain structural consistency between the new topology and the ideal topology.

[0029] This cascaded reconstruction scheme perfectly embodies the core ideas of "capability transfer" and "semantic topology self-healing." The result after replanning is as follows: Figure 4 As shown, U_E1, acting as a J-capability node, and U_S1, acting as an S-capability node, together with U_R2 and U_C1, reconstructed the semantic topology that satisfies the heterogeneous capability combination constraint.

[0030] Step 7: Instruction Distribution and Execution Commands to switch capability modes and new routes were distributed. While the swarm moved in physical space, it performed a "surgical" topology reconstruction in capability space.

[0031] This embodiment fully demonstrates the unique capabilities of the present invention in handling the decoupling of the physical layer and the capability layer, the constraint of heterogeneous capability combination, and the cascaded self-healing, which are not available in any traditional method.

Claims

1. A self-healing path planning method for heterogeneous UAV swarms based on capability space modeling and semantic topology preservation, characterized in that, Includes the following steps: Step S1: For each drone in the swarm, establish a task capability space model independent of its physical geometry model; the task capability space model is used to describe at least one virtual capability region radiating from the drone with itself as the spatial reference origin, and each virtual capability region is defined by capability type, range parameter, directional constraint and activation state; Step S2: Designate at least one key asset drone according to the mission scenario and construct a mission capability semantic topology graph; the semantic topology graph uses the virtual capability regions carried by each drone as semantic nodes and the effective coverage relationship of the virtual capability regions to the key asset drone or other drones as directed semantic edges; the effective coverage relationship is determined according to the coverage determination function in the corresponding capability space model. Step S3: Define the mission capability topology constraints for the key asset UAV. The mission capability topology constraints are directly applied to the attributes of the semantic nodes and directed semantic edges of the mission capability semantic topology graph. These constraints include the number of semantic edges covering specific capability types of the key asset, the spatial distribution angular dispersion of capability source nodes, and the logical combination relationship between semantic edges of different capability types. Step S4: Under the condition of satisfying the task capability topology constraints and flight performance constraints, plan the initial cooperative flight path for each drone in the swarm, so that each drone can carry and maintain the required semantic topology structure through physical movement. Step S5: Monitor the status of each node in the bee colony in real time. When a physical node failure is detected, causing the topological attributes of the task capability semantic topology graph to violate the task capability topology constraints, trigger semantic topology repair replanning. Step S6: Within the capability space of the task capability semantic topology graph, parse the missing semantic edge type of the current capability gap and its required spatial attributes, and generate an ideal replacement capability node description; query the corresponding task capability space model from the remaining UAVs, and select candidate UAVs whose potential capability list contains the required capability type and can be reconstructed in capability mode; with the goal of minimizing task performance loss and range cost, solve the joint optimization scheme containing capability mode switching instructions and new physical paths; The capability mode switching command enables the candidate UAV to activate its potential capabilities to generate new virtual capability regions, thereby creating new semantic nodes in the semantic topology graph, and using physical movement to fill capability gaps with the new semantic nodes, thus restoring the semantic topology structure. Step S7: Issue the capability mode switching command and path replanning command, and the bee colony executes the updated capability mode and physical path.

2. The method according to claim 1, characterized in that, In step S1, the task capability space model is defined as an extensible capability tuple: in, The physical location of the drone serves as the origin of its capability space; This is the set of currently activated ability types. To reconstruct the set of potential capability types that can be activated through pattern reconstruction; The virtual capability region is defined by the geometry spanned in space by the activated capability types and their coverage determination functions.

3. The method according to claim 3.2, characterized in that, In step S2, the construction of the directed semantic edge is based on the determination results of the physical locations of other nodes in the space by the coverage determination function set FiFi; if for the activated capability cc of the UAV uiui, its coverage determination function Then there exists a directed semantic edge from the capability semantic node carried by uiui to ujuj, and its edge type is capability type cc.

4. The method according to claim 1, characterized in that, In step S3, the logical combination relationship constraints between the different types of capability semantic edges include: requiring that the critical asset UAV be simultaneously pointed to by a directed semantic edge of type reconnaissance coverage and a directed semantic edge of type electronic jamming coverage, and that there must be a directed semantic path connection of type communication relay between the capability source nodes providing these two types of edges.

5. The method according to claim 1, characterized in that, In step S6, the process of parsing the missing semantic edge type of the current capability gap and its required spatial attributes specifically includes: calculating the existing coverage area of ​​each capability type of the key asset UAV in the capability space, comparing it with the task capability topology constraints, obtaining the missing coverage angle range and / or missing link for each required capability type, thereby generating the ideal replacement capability node description described by the target capability type, desired spatial orientation, and desired action distance.

6. The method according to claim 1 or 5, characterized in that, In step S6, the objective function of the joint optimization scheme is: in, Indicates candidate drones Whether to select the capability pattern refactoring To compensate for the performance loss caused by abandoning the current task. For the cost of the voyage, This is a penalty term for the structural deviation between the actual semantic topology and the ideal topology after the padding. The weighting coefficients are used; the solution process must satisfy the semantic topology recovery constraints and flight performance constraints after capability pattern reconstruction.

7. A self-healing path planning system for heterogeneous UAV swarms based on capability space modeling and semantic topology preservation, characterized in that, include: The capability space modeling module is used to build a mission capability space model for each UAV that is independent of the physical geometry model. This model includes the set of activated capabilities and potential capabilities and their coverage determination function. The semantic topology graph construction and maintenance module is used to build and update the mission capability semantic topology graph in real time, with each UAV as the physical carrier, virtual capability regions as semantic nodes, and capability coverage relationships as directed semantic edges. The topology constraint definition and monitoring module is used to define the task capability topology constraints imposed on the semantic topology graph and continuously verify whether the attributes of the semantic topology graph satisfy the constraints. The constraints include the number, spatial distribution, and logical combination of semantic edges of specific capability types covering key assets. The semantic topology repair solver is used to resolve capability gaps in the capability space when constraints are violated, filter reconfigurable nodes based on the potential capability list of each UAV, solve the joint optimization problem of capability mode switching and physical path planning, and generate a self-healing solution. The instruction distribution and synchronization module is used to distribute capability mode switching instructions and new physical paths to the corresponding UAVs for execution.

8. The system according to claim 7, characterized in that, When solving the joint optimization problem, the semantic topology repair solver adopts the objective function as described in claim 6 and satisfies the semantic topology recovery constraints after capability pattern reconstruction.

Images