A multi-level control mode adaptive switching method and system based on task constraints

Abstract

This invention discloses a multi-level control mode adaptive switching method and system based on task constraints. For different task types, the remote control center issues different task type identifiers and corresponding task constraint strategy information. During the task initiation preparation phase and task execution, the nest-side edge computing controller continuously collects network, task, and UAV status data, dynamically determines the most suitable target control mode, and performs control handover, state synchronization, and instruction semantic level switching when the target control mode is inconsistent with the current control mode. This achieves coordinated and consistent adjustment of control mode, control ownership, task state machine, and instruction semantics. This invention improves the continuity, security, and stability of UAV system task execution under different network conditions and task constraints.

Description

Technical Field

[0001] This invention relates to the field of automatic control of unmanned aerial vehicle (UAV) systems, and in particular to a method and system for adaptive switching of multi-level control modes based on task constraints. Background Technology

[0002] Existing UAV nesting systems typically employ a three-tier architecture: a remote control center, a UAV nest, and UAVs. The remote control center handles mission management, flight path planning, remote monitoring, data storage, and human interaction. The UAV nest is deployed at the mission site for UAV parking, charging or battery swapping, automatic hatch opening, takeoff and landing assistance, mission relay, and on-site monitoring. The UAVs, as the main entities performing the mission, are used to complete inspection, reconnaissance, mapping, or monitoring tasks.

[0003] During mission execution, the remote control center sends mission packages, advanced mission instructions, flight path documents, or control commands to the drone's control center via a wide area network. The control center then forwards the relevant mission information to the drone via a local wireless link and coordinates actions such as hatch opening, takeoff preparation, recovery and landing, and recharging. The content sent by the remote control center can be either low-level flight control commands or advanced mission commands that need to be broken down, translated, and locally planned by the control center. Video, images, and telemetry data collected by the drone during the mission are aggregated by the control center and transmitted back to the remote control center.

[0004] Currently, mainstream UAV nest products typically have certain edge computing capabilities and are equipped with state awareness modules, data storage units, communication modules, and mechanism control units. They can complete tasks such as task caching, flight path file reception, local path organization, task command translation, take-off and landing linkage control, telemetry data aggregation, data caching, and some edge processing tasks locally.

[0005] However, existing UAV nesting systems still have shortcomings in multi-level control mode coordination, control allocation, and adaptive mode switching, making it difficult to adapt to complex task constraints and dynamic network environments.

[0006] Problems and shortcomings of existing technologies

[0007] (1) Due to the lack of a mechanism for establishing differentiated control mode switching constraints based on task type in the existing technology, it is impossible to reasonably select the control mode according to the task requirements in different task scenarios, making it difficult to achieve unified judgment of task constraints, network status and edge resource status, thus affecting the adaptability of the system under multiple task types.

[0008] (2) Due to the lack of a stable hierarchical switching judgment mechanism in the existing technology under the link degradation scenario, when the wide area link experiences increased latency, increased packet loss rate, decreased bandwidth or increased jitter, the system can usually only make coarse-grained online or offline judgments, or simply adopt a single strategy such as continue execution, return to home or wait for recovery, resulting in frequent switching of control mode, unstable task execution status and chaotic instruction execution.

[0009] (3) Due to the lack of a hierarchical configuration mechanism for the main judgment right of task feasibility and the execution translation right in the existing system under different control modes, the decision-making responsibilities of the remote control center and the nest under different modes are not clearly defined. In particular, when the network status changes, there is no mechanism to smoothly transfer the main judgment right of task feasibility from the remote control center to the nest side, which can easily lead to overlapping control responsibilities or blurred boundaries, thereby affecting the stability and consistency of task execution.

[0010] (4) Because the existing technology lacks an effective task status synchronization and data retransmission mechanism after the link is restored, it only re-establishes the communication connection and continues data transmission after the network is restored. It lacks a mechanism for task status alignment, priority synchronization of key data, background retransmission of historical data, and re-determination of control mode, which can easily cause inconsistencies in status and control conflicts between the remote control center and the nest.

[0011] (5) Although the existing nested system has edge computing capabilities, it lacks a unified collaborative architecture around the multi-level control mode. The existing modules have not established a systematic coupling relationship around the task state machine, mode switching rules, control handover, migration of the main judgment right of task feasibility and instruction semantic switching, resulting in insufficient overall system collaboration capability. Summary of the Invention

[0012] Purpose of the Invention: To address the aforementioned problems, this invention provides a multi-level control mode adaptive switching method and system based on task constraints. For different task types, the remote control center issues different task type identifiers and corresponding task constraint strategy information. During the task startup preparation phase and task execution, the edge computing controller on the nest side continuously collects network, task, and UAV status data, dynamically determines the most suitable target control mode, and performs control handover, state synchronization, and instruction semantic level switching when the target control mode is inconsistent with the current control mode. This achieves coordinated and consistent adjustment of control mode, control ownership, task state machine, and instruction semantics, thereby ensuring the continuity, security, and stability of task execution under different network conditions and task constraints.

[0013] Technical solution: A multi-level control mode adaptive switching method based on task constraints, comprising the following steps:

[0014] Predefine multi-level control modes and configure task constraint policy library, mode determination rule library and mode determination trigger conditions;

[0015] The remote control center generates a task type identifier based on the task type, retrieves the corresponding task constraint policy information from the task constraint policy library, and sends it to the nest side; the nest side edge computing controller collects necessary status information during the task startup preparation phase and determines the initial control mode.

[0016] During mission execution, the nest continuously collects multi-source status information;

[0017] The edge computing controller determines whether the mode determination triggering condition is met based on the collected multi-source state information; if the condition is met, it initiates state evaluation and constructs a set of mode determination input parameters; if the condition is not met, it maintains the current control mode.

[0018] The edge computing controller determines a set of candidate control modes based on task constraint policy information. Based on the set of mode determination input parameters, it performs a state feasibility determination on the candidate control modes according to the mode determination rule base to determine the target control mode. If the target control mode is different from the current control mode, it performs control handover, state synchronization, and instruction semantic level switching. If there is no candidate control mode that meets the conditions, it performs a backup action.

[0019] Before the edge computing controller performs a mode switch check, it freezes the current instruction queue and generates a task state snapshot, and updates the main control entity identifier, the task state machine maintenance entity, the main entity with the main authority to judge task feasibility, the control instruction issuance channel, the instruction parsing logic, and the instruction semantic level. After the mode switch is completed, the instruction execution queue is restored to ensure that only one valid main control entity exists in the system at any given time.

[0020] Furthermore, the multi-level control modes include transparent transmission mode, intelligent relay mode, edge autonomous mode, and independent master control mode. Each mode is distinguished based on the task state machine maintaining entity, the attribution of the main judgment authority for task feasibility, and the semantic hierarchy of control commands. The task constraint policy library stores task constraint policy information corresponding to each task type, including the set of control modes that can be switched, the priority of the target control mode, and the backup action. The mode determination rule library is configured on the nest side, defining the entry conditions, maintenance conditions, and exit conditions for each control mode. The multi-source state information includes network link state information, task state information, edge resource state information, and UAV operating state information.

[0021] Furthermore, in the transparent transmission mode, the remote control center acts as the main control entity, directly issuing low-level flight control commands to the UAV; in the intelligent relay mode, the remote control center acts as the main control entity, and the UAV's nest performs task commands issued by the remote control center on a type-based basis, including splitting, translating, and protective verification; in the edge autonomy mode, the UAV's nest acts as the main control entity, and the commands from the remote control center serve as target inputs or intervention inputs; in the independent master control mode, the UAV's nest independently controls the UAV based on local data when the wide-area link cannot meet the remote control requirements.

[0022] The four control modes are distinguished based on the main body of task state machine maintenance, the ownership of the main judgment authority for task feasibility, the semantic level of control instructions, the depth of the control loop participation of the nest, and the ability of the system to continue to execute tasks after leaving the remote control center under link abnormal conditions.

[0023] In transparent transmission mode, the remote control center acts as the primary control entity, responsible for maintaining the mission state machine and possessing the primary authority to determine mission feasibility. It directly issues low-level flight control commands or fine-grained real-time control commands to the UAV. The drone's nest primarily handles communication transmission, status feedback, and takeoff and landing assistance functions, without participating in mission decision-making during execution. The control loop mainly resides between the remote control center and the UAV. This mode lacks the ability to execute missions independently of the remote control center and is suitable for mission scenarios requiring real-time human intervention and precise control.

[0024] In intelligent relay mode, the remote control center acts as the primary control entity, responsible for maintaining the mission state machine and possessing the primary authority to determine mission feasibility. The nacelle participates in the control loop, responsible for task decomposition, local path organization, protocol conversion, and flight control command decomposition of high-level mission instructions issued by the remote control center, and performing execution-level protective checks. Instructions issued by the remote control center are mission-level or action-level instructions, which are translated into flight control execution instructions at the nacelle. In this mode, the system still relies on the continuous participation of the remote control center and does not have the ability to complete tasks independently without its intervention.

[0025] In edge autonomy mode, the nest, as the primary control entity, is responsible for maintaining the task state machine and possesses the primary authority to determine task feasibility, thus leading the task execution closed loop. The remote control center transforms from a master node into a supervisory and intervention node. Its issued business commands are no longer directly used as the basis for execution but rather as high-level target inputs, constraint inputs, priority adjustment inputs, or intervention inputs. The nest, combining local state, task constraint strategies, and local control rules, determines task feasibility and decides whether, when, and how to execute the task. In this mode, the control closed loop primarily resides between the nest and the drone, allowing the system to continue executing tasks to a certain extent independently of the remote control center.

[0026] In independent master control mode, when the wide area link is interrupted or cannot meet remote control requirements, the nest acts as the sole master control entity, maintaining the task state machine and possessing the primary authority to determine task feasibility. Based on local task templates, cached task contexts, local map data, and local control rules, the nest independently completes task execution, exception handling, return-to-home, and recovery control. In this mode, the system operates completely independently of the remote control center, possessing the capability to execute tasks even under link failure conditions.

[0027] The task constraint policy library, configured in the remote control center, is a collection used to store and manage multiple task constraint policy information entries. Each task constraint policy entry is indexed and uniquely associated using a task type identifier. This library supports pre-defining differentiated task-side constraint rules for different tasks based on task type. The task constraint policy information includes at least the following constraint fields: the set of allowed control modes, the set of prohibited control modes, whether continued execution is allowed after a link failure, whether remote continuous decision-making is required, whether local takeover is allowed, the allowed instruction semantic level during mode switching, the initial attribution and transfer authority for task feasibility assessment, the backup action in case of link deterioration or no feasible mode, the allowed flight phases, the prohibited flight phases, and the priority of the target control mode.

[0028] A mode determination rule base, configured on the edge computing controller at the nest side, is used to define the entry, maintenance, and exit conditions for each control mode from the perspective of the system's real-time status. In the mode determination rule base, each control mode corresponds to a rule template, which includes at least: mode identifier, entry conditions, maintenance conditions, exit conditions, link quality requirements, edge resource requirements, local task template integrity requirements, local map data integrity requirements, local task context integrity requirements, local control rule validity requirements, UAV safety status requirements, task phase requirements, task state machine maintenance entity, attribution of primary judgment authority for task feasibility, control command semantic level, and exception handling rules.

[0029] Furthermore, the mode determination triggering conditions are used to determine when to start the target control mode determination process, including: network link triggering conditions, edge resource triggering conditions, task phase triggering conditions, task constraint triggering conditions, UAV operating status triggering conditions, and periodic scheduling triggering conditions.

[0030] Furthermore, the set of input parameters for pattern determination includes multi-source state information and the results of state assessment, which includes network link state assessment, edge resource availability assessment, UAV operation state assessment, and mission state assessment.

[0031] Furthermore, the network link state assessment includes constructing a comprehensive network quality index. , used to characterize the support degree of the current link for remote control capabilities;

[0032] ;

[0033] Among them, is the delay normalization function, is the packet loss rate, is the bandwidth normalization function, is the jitter normalization function, is the weight coefficient;

[0034] The evaluation of the edge resource availability includes evaluating the processing capacity, storage capacity, integrity of local task templates, integrity of local map data, integrity of local task context, validity of local control rules, status of the airframe execution mechanism, and status of the airframe energy; it is used to determine whether the airframe side has the ability to execute autonomously or control independently;

[0035] The evaluation of the UAV operation status includes evaluating the power status, position, speed, attitude, payload status, obstacle avoidance status, etc., and is used to evaluate whether the UAV itself is in a safe state suitable for switching;

[0036] The evaluation of the task status includes evaluating the current task stage, task progress, whether remote continuous decision-making is still required currently, and whether the conditions for offline continued execution are met, and is used to provide the status input of the task side for the subsequent screening of candidate control modes.

[0037] Furthermore, the decision rules for the state feasibility determination include:

[0038] Entry conditions for the transparent transparent transmission mode: ≥T1, and the remote control center can continuously obtain complete status information; T1 is the link quality threshold for supporting the transparent transparent transmission mode;

[0039] Entry conditions for the intelligent relay mode: T2≤ <T1, the remote control center can continuously obtain complete status information, and the airframe has the capabilities of task splitting, protocol conversion, and instruction translation; T2 is the link quality threshold for supporting the intelligent relay mode;

[0040] Entry conditions for the edge autonomous mode: <T2 and the wide-area link has not been completely interrupted, and the edge resource status, local task template, local map data, and task context meet the autonomous execution conditions;

[0041] Entry conditions for the independent main control mode: The continuous interruption duration of the wide-area link exceeds the preset threshold T3, and the edge resource status, local task template, task context, local control rules, and UAV operation status meet the independent execution conditions;

[0042] When multiple candidate control modes meet the feasibility conditions, the system determines the most suitable target control mode according to the priority of the target control mode, the current task stage, the task continuity requirements, and the security requirements in the task constraint strategy information.

[0043] When the target control mode is consistent with the current control mode, the system maintains the current control mode and continues to execute tasks and monitor the status. When the target control mode is inconsistent with the current control mode, the system enters the control mode switching process, namely, the handover of control, status synchronization, and instruction semantic level switching. When the target control mode switching conditions are detected, the current status is required to continue for a preset hysteresis time before the control mode switching and status synchronization are executed.

[0044] When there is no candidate control mode that meets the conditions, or when a safety event such as low battery, obstacle avoidance trigger, abnormal positioning, or abnormal attitude occurs, causing the current task to be unable to continue to be executed according to any candidate control mode, the system executes the preset backup action in the task constraint strategy information. The backup action includes one or more of the following: hovering and waiting, returning to home, landing, continuing the current safety action, waiting for manual takeover, or terminating the task.

[0045] Furthermore, during the mode switching process, for cases where the target control mode is transparent transmission mode, intelligent relay mode, or edge autonomous mode, a consistency comparison is performed between the local task status of the data center and the task status recorded by the remote control center; for cases where the target control mode is independent master control mode, the consistency comparison is skipped, and the switching events and task progress are synchronized through background retransmission after the link is restored.

[0046] A task-constrained multi-level control mode adaptive switching system is used for dynamic switching of control modes between UAV nest and remote control center. It includes: edge computing controller, multi-link communication module, state perception module, data storage unit, energy management unit, nest execution unit, UAV and remote control center.

[0047] The edge computing controller is the core control unit on the nest side, used to execute the steps of a multi-level control mode adaptive switching method based on task constraints, realize the adaptive switching of control mode and dynamic handover of control, and parse and execute control commands according to the current control mode.

[0048] The multi-link communication module is used to establish a local wireless communication link between the nest and the UAV, as well as a wide-area communication link between the nest and the remote control center, and to transmit control commands, telemetry data, mission status data and payload data.

[0049] The state awareness module is used to collect multi-source state information related to task execution and send it to the edge computing controller; the multi-source state information includes network link status, nest device status, edge resource status, and UAV operating status;

[0050] The data storage unit is used to store task templates, flight path files, map data, no-fly zone data, task breakpoint information, recovery parameters, and telemetry data, video data, and log data generated during task execution, in order to support task execution, mode switching, and disconnection recovery control.

[0051] The energy management unit is used to manage the power supply status of the nest and the UAV, and to provide power supply status, charging status and battery status information to the edge computing controller for mission take-off determination, endurance assessment and return triggering.

[0052] The nest execution unit is used to execute nest mechanism actions under the control of the edge computing controller to cooperate with the UAV to complete the take-off, landing, recovery, reset and recharging processes;

[0053] The remote control center is used to create tasks, generate task packages, issue control commands, and remotely supervise and manage the task execution process. Specifically, the remote control center is used to generate task type identifiers, task content, and task constraint information, and distribute them to the nest side via a wide area network. At the same time, it receives telemetry data, payload data, and task status information transmitted back from the nest side and the UAV to support task monitoring and anomaly handling.

[0054] Furthermore, the edge computing controller is connected to the multi-link communication module, the state awareness module, the data storage unit, the energy management unit, and the nest execution unit, respectively; the multi-link communication module is connected to the UAV and the remote control center to realize local wireless communication and wide area network communication.

[0055] Beneficial effects: (1) Realize differentiated control mode constraints based on task type, and improve the system's adaptability to multi-task scenarios. By introducing task type identifiers and task constraint strategy library, the initial control mode and switchable boundary are determined in the task initialization stage, so that different tasks can select appropriate control modes according to preset constraints during execution, avoiding the problem of mismatch between control mode selection and task requirements in the prior art, thereby improving the rationality and reliability of task execution.

[0056] (2) Achieve stable switching of control modes based on multi-source states and improve the system's operational stability in complex network environments. By jointly evaluating network link status, edge resource status, UAV operation status and mission phase, and selecting control modes based on unified judgment rules, while introducing a hysteresis mechanism, the system can effectively avoid frequent switching and oscillation problems of control modes, thereby improving the system's stability in scenarios with link fluctuations or degradation.

[0057] (3) To achieve consistency between control authority and task status during control mode switching and avoid control conflicts. During control mode switching, the task state machine maintenance subject and the main judgment authority of task feasibility are synchronously transferred through the control handover mechanism, and the instruction execution queue is frozen and restored to ensure that there is only one main control subject at any time, thereby avoiding the problem of instruction conflict and execution chaos caused by unclear control subject in the prior art.

[0058] (4) Achieve adaptive switching of the semantic level of control commands to improve the rationality of system execution under different control modes. By dynamically adjusting the semantic level and activation method of remote control center commands under different control modes, the same type of business commands have different processing logic and execution paths under different modes, thereby ensuring that the semantics of commands are consistent and interpretable during the switching between remote master control and local master control, and improving the controllability and consistency of system control behavior.

[0059] (5) Achieve continuous task execution capability under link degradation and interruption scenarios to improve system robustness. When the network link deteriorates significantly or is interrupted, the main decision-making power of task feasibility is transferred to the nest side, and the task execution loop is maintained by relying on local task templates and cached task context in independent master control mode. This enables the system to continue to complete tasks or safely terminate tasks without the remote control center, thereby improving the reliability of the system in abnormal environments.

[0060] (6) Realize task status synchronization and data retransmission after network recovery to ensure task status consistency. After the link is restored, through task status alignment, priority synchronization of key data, and background retransmission of historical data, the remote control center can regain a complete and accurate task status view, avoiding the inconsistency of status and control chaos caused by direct connection restoration in the existing technology.

[0061] (7) Construct a unified multi-level control and coordination architecture to improve the overall coordination efficiency of the system. By establishing a unified coordination architecture around the task state machine, mode switching rules, control handover mechanism, task feasibility master judgment transfer mechanism and instruction semantic switching mechanism, an organic coordination relationship is formed between the functional modules of the machine nest, thereby improving the overall control efficiency and resource utilization efficiency of the system. Attached Figure Description

[0062] Figure 1 This is a schematic diagram of the multi-level control mode adaptive switching system of the present invention;

[0063] Figure 2 This is a schematic diagram comparing the distribution of control subjects and judgment rights under the four control modes of this invention;

[0064] Figure 3 This is a flowchart of the multi-level control mode adaptive switching method of the present invention;

[0065] Figure 4 This is the control mode switching determination logic diagram of the present invention;

[0066] Figure 5 This is a schematic diagram of the control transfer process of this invention. Detailed Implementation

[0067] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0068] This invention provides a multi-level control mode adaptive switching method based on task constraints, comprising the following steps:

[0069] Step 1: Pre-configure multi-level control modes and rule system.

[0070] Before task execution, the system predefines multiple control modes and configures a task constraint strategy library, a mode determination rule library, and mode determination triggering conditions. This invention includes at least a transparent transmission mode, an intelligent relay mode, an edge autonomous mode, and an independent master control mode. These four control modes are differentiated based on the task state machine maintaining entity, the attribution of the main judgment authority for task feasibility, the semantic level of control commands, the depth of the control loop involving the homing system, and the system's ability to continue executing tasks even when disconnected from the remote control center due to link anomalies.

[0071] In transparent transmission mode, the remote control center acts as the primary control entity, responsible for maintaining the mission state machine and possessing the primary authority to determine mission feasibility. It directly issues low-level flight control commands or fine-grained real-time control commands to the UAV. The drone's nest primarily handles communication transmission, status feedback, and takeoff and landing assistance functions, without participating in mission decision-making during execution. The control loop mainly resides between the remote control center and the UAV. This mode lacks the ability to execute missions independently of the remote control center and is suitable for mission scenarios requiring real-time human intervention and precise control.

[0072] In intelligent relay mode, the remote control center acts as the primary control entity, responsible for maintaining the mission state machine and possessing the primary authority to determine mission feasibility. The nacelle participates in the control loop, responsible for task decomposition, local path organization, protocol conversion, and flight control command decomposition of high-level mission instructions issued by the remote control center, and performing execution-level protective checks. Instructions issued by the remote control center are mission-level or action-level instructions, which are translated into flight control execution instructions at the nacelle. In this mode, the system still relies on the continuous participation of the remote control center and does not have the ability to complete tasks independently without its intervention.

[0073] In edge autonomy mode, the nest, as the primary control entity, is responsible for maintaining the task state machine and possesses the primary authority to determine task feasibility, thus leading the task execution closed loop. The remote control center transforms from a master node into a supervisory and intervention node. Its issued business commands are no longer directly used as the basis for execution but rather as high-level target inputs, constraint inputs, priority adjustment inputs, or intervention inputs. The nest, combining local state, task constraint strategies, and local control rules, determines task feasibility and decides whether, when, and how to execute the task. In this mode, the control closed loop primarily resides between the nest and the drone, allowing the system to continue executing tasks to a certain extent independently of the remote control center.

[0074] In independent master control mode, when the wide area link is interrupted or cannot meet remote control requirements, the nest acts as the sole master control entity, maintaining the task state machine and possessing the primary authority to determine task feasibility. Based on local task templates, cached task contexts, local map data, and local control rules, the nest independently completes task execution, exception handling, return-to-home, and recovery control. In this mode, the system operates completely independently of the remote control center, possessing the capability to execute tasks even under link failure conditions.

[0075] The task constraint policy library, configured in the remote control center, is a collection used to store and manage multiple task constraint policy information entries. Each task constraint policy entry is indexed and uniquely associated using a task type identifier. This library supports pre-defining differentiated task-side constraint rules for different tasks based on task type. The task constraint policy information includes at least the following constraint fields: the set of allowed control modes, the set of prohibited control modes, whether continued execution is allowed after a link failure, whether remote continuous decision-making is required, whether local takeover is allowed, the allowed instruction semantic level during mode switching, the initial attribution and transfer authority for task feasibility assessment, the backup action in case of link deterioration or no feasible mode, the allowed flight phases, the prohibited flight phases, and the priority of the target control mode.

[0076] A mode determination rule base, configured on the edge computing controller at the nest side, is used to define the entry, maintenance, and exit conditions for each control mode from the perspective of real-time system status. In the mode determination rule base, each control mode corresponds to a rule template, which includes at least: mode identifier, entry conditions, maintenance conditions, exit conditions, link quality requirements, edge resource requirements, local task template integrity requirements, local map data integrity requirements, local task context integrity requirements, local control rule validity requirements, UAV safety status requirements, task phase requirements, task state machine maintenance entity, task feasibility master judgment authority, control command semantic level, and exception handling rules.

[0077] The mode determination trigger conditions are used to determine when to initiate the target control mode determination process. These trigger conditions include at least network link trigger conditions, edge resource trigger conditions, mission phase trigger conditions, mission constraint trigger conditions, UAV operational status trigger conditions, and periodic scheduling trigger conditions. Trigger conditions are set based on state changes.

[0078] Step 2: Determine the initial control mode of the task.

[0079] During the task creation phase, the remote control center determines the task type based on task requirements and generates a task type identifier. Based on this identifier, the remote control center retrieves the corresponding task constraint policy information from the task constraint policy library and sends the task type identifier and task constraint policy information along with the task package to the drone's nest side. During the task startup preparation phase, the nest-side edge computing controller powers on the drone and collects necessary state information before task startup. This necessary state information includes at least network link status, edge resource status, drone initial status, nest energy status, nest actuator status, local task template integrity, local map data integrity, and local task context integrity. The edge computing controller treats task startup preparation as a mode determination trigger event and proceeds to subsequent state evaluation, candidate control mode screening, and target control mode selection processes to determine the initial control mode for formal task execution.

[0080] Step 3: Continuous acquisition of multi-source status.

[0081] During mission execution, the nest continuously collects network link status information, mission status information, edge resource status information, and UAV operational status information. The network link status information includes latency, packet loss rate, available bandwidth, jitter, duration of consecutive link outages, and recovery duration for both wide-area and local links. The mission status information includes mission stage, mission progress, whether it is in a flight stage where switching is permitted, whether remote continuous decision-making is still required, and whether offline continuation conditions are met. The edge resource status information includes edge computing controller processor utilization, memory utilization, remaining storage space, local map data integrity, local mission template integrity, nest energy status, nest actuator health status, and takeoff and landing auxiliary mechanism status. The UAV operational status information includes position, speed, attitude, battery level, payload status, flight path execution status, obstacle avoidance status, and landing preparation status.

[0082] Step 4: Pattern Determination Trigger Detection

[0083] The edge computing controller determines whether any of the triggering conditions in the pattern determination triggering conditions is met based on continuously collected multi-source state information. The triggering conditions include at least:

[0084] Network link triggering conditions include: the comprehensive network quality index crossing a preset threshold, the wide area link being interrupted or restored, the continuous interruption duration of the wide area link reaching a preset threshold, the link quality failing to meet the maintenance conditions of the current control mode, and the remote control center being unable to continuously obtain complete status information.

[0085] Edge resource triggering conditions include edge computing controller processor utilization, memory utilization, remaining storage space, local map data integrity, local task template integrity, local task context integrity, local control rule validity, and the nest energy status or nest actuator status not meeting the current control mode maintenance conditions.

[0086] The mission phase triggering conditions include changes in the mission phase, or the current mission entering or exiting the mission constraint policy that allows or prohibits switching flight phases.

[0087] Task constraint triggering conditions include changes in task constraint policy information or task constraint policy library version, which cause changes in the priority of allowed control mode, prohibited control mode, local takeover permission, disconnected execution permission, remote continuous decision-making requirements, backup action or target control mode.

[0088] The conditions that trigger the drone's operational status include low battery, obstacle avoidance trigger, abnormal positioning, abnormal attitude, deviation from the flight path, abnormal speed, abnormal payload, abnormal landing preparation status, or abnormal local communication.

[0089] Periodic scheduling trigger conditions include the edge computing controller performing current control mode maintenance condition checks and target control mode re-determination according to a preset period.

[0090] When any of the above triggering conditions are met, the edge computing controller initiates the state assessment and judgment input construction process; when the triggering conditions are not met, the system maintains the current control mode and continues to perform state acquisition and triggering condition detection.

[0091] Step 5: State Evaluation and Judgment Input Construction

[0092] When the pattern determination triggering condition is met, the edge computing controller performs a unified quantitative evaluation of the current system operating status based on the latest collected multi-source status information, and constructs a set of pattern determination input parameters.

[0093] (1) Network link state assessment

[0094] To assess network link status, a comprehensive network quality index is constructed. This metric is used to characterize the degree to which the current link supports remote control capabilities. The metric can be expressed as:

[0095]

[0096] in, For link latency, For packet loss rate, For available bandwidth, For link jitter; , , , These are the normalization functions for latency, packet loss rate, bandwidth, and jitter, respectively. These are the corresponding weighting coefficients.

[0097] (2) Multi-source state joint assessment

[0098] In addition to network status, the edge computing controller also extracts and confirms key status information from the following dimensions as input parameters for pattern determination:

[0099] Edge resource availability includes processing power, storage capacity, local task template integrity, local map data integrity, local task context integrity, local control rule validity, nest actuator status, and nest energy status, used to determine whether the nest side has autonomous execution or independent control capabilities.

[0100] Drone operating status: including battery status, position, speed, attitude, payload status, obstacle avoidance status, etc., used to assess whether the drone itself is in a safe state suitable for switching.

[0101] Task status: This includes the current task stage, task progress, whether remote continuous decision-making is still required, and whether the conditions for offline continuation are met. It provides task-side status input for subsequent candidate control mode selection.

[0102] The network evaluation results mentioned above, together with the state information of each dimension, constitute the set of input parameters for control mode determination, which are used to determine the feasibility of subsequent candidate control modes and select the target control mode.

[0103] Step Six: Candidate Control Mode Screening and Target Control Mode Selection

[0104] After forming the set of input parameters for pattern determination, the edge computing controller first determines the set of candidate control modes allowed for the current task based on the task constraint policy information. If the current task phase is a prohibited switching phase, the candidate control mode set only retains the current control mode or the backup action.

[0105] After completing the candidate control mode screening, the edge computing controller performs a state feasibility determination on each candidate control mode based on the mode determination input parameter set. The state feasibility determination is performed according to the entry conditions defined in the mode determination rule base. In one implementation, the specific determination rules include:

[0106] Conditions for entering transparent pass-through mode: Furthermore, the remote control center can continuously obtain complete status information;

[0107] Conditions for entering intelligent relay mode: The remote control center can continuously obtain complete status information, and the nest has the ability to split tasks, convert protocols, and translate instructions.

[0108] Entry conditions for edge autonomy mode: Furthermore, the wide area link has not been completely interrupted, and the edge resource status, local task template, local map data, and task context meet the conditions for autonomous execution.

[0109] Entry conditions for independent master control mode: The duration of continuous wide area link interruption exceeds a preset threshold. The edge resource status, local task template, task context, local control rules, and drone operating status meet the conditions for independent execution.

[0110] When multiple candidate control modes meet the feasibility conditions, the system determines the most suitable target control mode based on the priority of the target control mode, the current task stage, the task continuity requirements, and the security requirements in the task constraint strategy information.

[0111] When the target control mode is consistent with the current control mode, the system maintains the current control mode and continues to execute tasks and monitor status. When the target control mode is inconsistent with the current control mode, the system enters a control mode switching process, namely, control handover, status synchronization, and instruction semantic level switching.

[0112] When there is no candidate control mode that meets the conditions, or when a safety event such as low battery, obstacle avoidance trigger, abnormal positioning, or abnormal attitude occurs, causing the current task to be unable to continue to be executed according to any candidate control mode, the system executes the preset backup action in the task constraint strategy information. The backup action includes one or more of the following: hovering and waiting, returning to home, landing, continuing the current safety action, waiting for manual takeover, or terminating the task.

[0113] During the continued execution of the task, the edge computing controller continuously performs multi-source state acquisition and mode determination trigger detection; when the new mode determination trigger condition is met, the system re-executes the state evaluation, candidate control mode screening and target control mode selection process, thereby forming a closed loop of dynamic adjustment of control mode.

[0114] To prevent frequent oscillations, the system incorporates a hysteresis mechanism during the target control mode determination process. When the target control mode switching conditions are detected, the system requires that the state remain in that state for a preset hysteresis time before executing the control mode switch and state synchronization.

[0115] Step 7: Control mode switching execution and state synchronization

[0116] When the target control mode differs from the current control mode, the edge computing controller executes a control handover, state synchronization, and instruction semantic level switching process. Depending on whether the target control mode depends on a remote control center, the process is divided into a complete handover process and a local independent handover process. The process includes:

[0117] First, the system performs a pre-switch verification of the current mission phase, the UAV's safety status, the conditions for entering the target control mode, and the availability of the target's main control entity. If the pre-switch verification fails, the system does not execute the subsequent control handover process and maintains the current control mode or executes a preset backup action based on the mission constraint policy information.

[0118] Secondly, the current queue of pending instructions is frozen, new process control instructions are suspended, and a task status snapshot is generated. The task status snapshot includes the current control mode, target control mode, current main control entity, target main control entity, task state machine status, task progress, current waypoint or work point, UAV position, speed, attitude, battery level, payload status, alarm status, key control parameters, ownership of current task feasibility assessment authority, and current instruction semantic level.

[0119] Then, for cases where the target control mode is transparent transmission mode, intelligent relay mode or edge autonomous mode, alarm information, key telemetry information, task progress information and task completion status are synchronized first, and the consistency of the task status in the local nest and the task status recorded by the remote control center is compared.

[0120] Once the consistency comparison passes, the main control subject identifier, task state machine maintenance subject, task feasibility main judgment subject, control command issuance channel, command parsing logic, and command semantic level are updated according to the target control mode.

[0121] In the case where the target control mode is the independent master control mode, since the wide area link has been interrupted, the priority synchronization and consistency comparison steps are skipped. Without the participation of the remote control center, the edge computing controller directly performs local updates of control attributes based on the local task status snapshot, and synchronizes the switching events and task progress to the remote control center through background retransmission after the link is restored.

[0122] Subsequently, after confirming that the target main control entity has completed the state takeover, the instruction execution queue is restored, and the task continues to be executed according to the control rules corresponding to the target control mode;

[0123] For historical videos, ordinary operation logs, non-critical load data, and non-real-time status data, the system synchronizes them to the remote control center through background retransmission after the control mode switch is completed, without blocking the real-time control mode switch and control handover.

[0124] When the task status consistency comparison fails, or when anomalies such as status synchronization failure, target main control subject takeover timeout, instruction semantic level switching failure, or instruction queue freezing failure occur, the system suspends the current control handover process and executes the status conflict handling strategy, control rollback, maintains the current security control state, or preset backup actions according to the task constraint policy information.

[0125] The above handover mechanism ensures that at any given time, there is only one valid master control entity in the system, thereby avoiding control conflicts, duplicate execution of instructions, or inconsistent task states.

[0126] When switching control modes, the system synchronously adjusts the semantic level of control commands allowed by the remote control center, as well as the corresponding parsing and activation methods.

[0127] In transparent transmission mode, the remote control center can use the underlying flight control command set, which includes speed, attitude, heading, altitude, or position control commands.

[0128] In intelligent relay mode, the remote control center uses advanced task instruction sets, task-level instruction sets, or action-level instruction sets. These instructions are executed on the nest side after task splitting, local path organization, protocol conversion, and flight control instruction decomposition.

[0129] In edge autonomy mode, the business commands issued by the remote control center can be the same as or similar to those in intelligent relay mode in terms of command name or surface task description. However, the parsing semantics, processing flow and activation method on the nest side are switched to high-level target input, constraint input, priority adjustment input, forced abort input or forced return input. The nest combines local status, local task constraint strategy and mode judgment rules to judge the feasibility of the task and decide whether to execute, when to execute and how to execute.

[0130] In independent master control mode, the machine nest is independently controlled based on local task templates, cached task contexts and local control rules, and the remote control center usually no longer issues process control commands in real time.

[0131] The main difference between the intelligent relay mode and the edge autonomous mode lies not in whether the command names are the same, but in the ownership of the primary judgment authority for task feasibility, control semantics, processing flow, and activation method. In the intelligent relay mode, similar business commands are process control commands after the remote control center completes the primary judgment of task feasibility, and the nest mainly performs instruction translation, local planning, and execution-level protective verification. In the edge autonomous mode, similar business commands are high-level target inputs or intervention inputs to the nest's local main control closed loop. The remote control center can make upper-level pre-judgments based on the received status, but the nest has the primary judgment authority for task feasibility and decides whether to execute and when to execute.

[0132] like Figure 1 As shown, the present invention also provides a multi-level control mode adaptive switching system 101 based on task constraints, including an edge computing controller 102, a multi-link communication module 103, a state perception module 104, a data storage unit 105, an energy management unit 106, a nest execution unit 107, a drone 108, and a remote control center 109.

[0133] The edge computing controller 102 is the core control unit on the nest side. It is used to receive task information and control commands issued by the remote control center 109, and control and manage the task execution process of the UAV 108 in combination with local status information.

[0134] Specifically, the edge computing controller 102 is used to determine the initial control mode and mode switching conditions according to the task type and task constraint strategy, and to determine and switch the control mode based on the network link status, edge resource status, UAV operation status and task stage during task execution; at the same time, during the mode switching process, the edge computing controller 102 is used to perform control handover, update the task state machine to maintain the subject and the subject of the task feasibility judgment, and parse, translate and execute control commands according to the current control mode.

[0135] In different control modes, the edge computing controller 102 undertakes functions such as instruction translation, local task execution control, or independent task closed-loop control.

[0136] Preferably, the task constraint management function, the task feasibility master judgment right switching management function, and the instruction semantic switching management function can be integrated into the edge computing controller 102.

[0137] The multi-link communication module 103 is used to establish a local wireless communication link between the nest and the UAV 108, and a wide-area communication link between the nest and the remote control center 109, and to transmit control commands, telemetry data, mission status data and payload data.

[0138] The state awareness module 104 is used to collect multi-source state information related to task execution and send it to the edge computing controller 102. The state information includes at least network link status, nest device status, edge resource status, and UAV operating status.

[0139] The data storage unit 105 is used to store task templates, flight path files, map data, no-fly zone data, task breakpoint information, recovery parameters, and telemetry data, video data and log data generated during task execution, in order to support task execution, mode switching and link failure continuation control.

[0140] The energy management unit 106 manages the power supply status of the nest and the drone 108, and provides power supply status, charging status and battery status information to the edge computing controller 102 for use in mission take-off determination, endurance assessment and return triggering.

[0141] The nest actuator 107 is used to perform nest mechanism actions under the control of the edge computing controller 102 to cooperate with the UAV 108 in completing the takeoff, landing, recovery, reset, and recharging processes. Specifically, this includes opening and closing the hatch, releasing and resetting the centering mechanism, performing actions of the auxiliary recovery mechanism, and performing actions of the recharging mechanism.

[0142] The UAV 108 is used to execute flight and payload tasks issued by the remote control center 109 or the edge computing controller 102, and transmits telemetry data, payload data and mission status information back during the execution of the mission.

[0143] The remote control center 109 is used to create tasks, generate task packages, issue control commands, and remotely monitor and manage the task execution process.

[0144] Specifically, the remote control center 109 is used to generate task type identifiers, task content and task constraint information, and distribute them to the nest side via a wide area network; at the same time, it receives telemetry data, payload data and task status information transmitted back from the nest side and the UAV 108 to support task monitoring and anomaly handling.

[0145] Example 1: Real-time manual fine control in transparent transmission mode

[0146] 1. Task creation and constraint loading

[0147] This embodiment uses a post-disaster site reconnaissance mission as an example. Such missions are characterized by uncertain targets, rapidly changing environments, and a high demand for real-time human judgment. They typically require remote operators to continuously perform fine-grained control based on on-site video footage, telemetry information, and unforeseen circumstances. Therefore, during the mission creation phase, the remote control center 109 determines the mission to be a post-disaster site reconnaissance mission based on the mission requirements and generates a corresponding mission type identifier. The remote control center 109 then retrieves the corresponding mission constraint policy information from the mission constraint policy library based on the mission type identifier. This policy information restricts the mission to require remote real-time participation and prohibits local decision-making.

[0148] 2. Determining the initial control state

[0149] Based on the aforementioned task type and constraints, the edge computing controller 102 powers on the UAV and collects necessary state information before the task starts. The edge computing controller 102 treats the task start preparation as a mode determination trigger event. After state evaluation and mode determination, it determines that the initial control mode of the task is the transparent pass-through mode.

[0150] 3. Task execution process

[0151] After the mission begins, the remote control center 109 continuously receives video transmissions, telemetry data, and nest status information from the UAV 108 via the multi-link communication module 103. Remote operators issue flight control and payload control commands in real time based on the on-site footage, such as attitude adjustment, speed adjustment, direction correction, hovering, minute displacement, gimbal rotation, fixed-point observation, and local magnification observation. After these commands are sent to the nest via a wide-area link, the edge computing controller 102 does not break down, replan, or make autonomous decisions regarding the mission content. Instead, it primarily transmits the underlying flight control commands or fine-grained real-time control commands to the UAV 108 via the multi-link communication module 103, while simultaneously handling status feedback, link maintenance, and takeoff and landing assistance.

[0152] 4. Control logic in transparent transmission mode

[0153] In this embodiment, the task execution state machine is maintained by the remote control center 109, which also undertakes the main judgment of task feasibility and real-time operation decisions. The nest side mainly undertakes communication pass-through, protocol adaptation, data aggregation, image transmission and feedback, and take-off and landing assistance functions. For example, when a remote operator controls the drone 108 to move 2 meters to the left, lower its height by 1 meter, and adjust the gimbal tilt angle for close-range shooting based on the damage to the building facade after a disaster, the edge computing controller 102 does not perform a task-level judgment on whether the action should be executed. Instead, it quickly passes through the corresponding control command to the drone 108 for execution and synchronously transmits the execution status and payload data back.

[0154] Preferably, in the transparent transmission mode, the edge computing controller 102 can still retain bottom-line protection functions to prevent obviously illegal or unissued control commands from directly entering the flight control link. For example, when receiving an incorrectly formatted command, a link verification failure command, or an abnormal command that is obviously beyond the capabilities of the local interface, the edge computing controller 102 can refuse to transmit the command and report the cause of the abnormality to the remote control center 109; however, this type of processing is a link protection verification and does not change the basic control relationship in the transparent transmission mode, in which the remote control center 109 assumes the primary responsibility for judging the feasibility of the task and the real-time control.

[0155] 5. Conditions for mode retention or subsequent switching

[0156] During task execution, if the state awareness module 104 detects that the wide area link remains excellent and the remote control center 109 remains online and can stably receive the field status, the system maintains the transparent transmission mode. If the quality of the wide area link is subsequently detected to be degraded, making it difficult to maintain fine-grained real-time control stably, the edge computing controller 102 will re-execute the state assessment and mode determination process when the mode determination trigger condition is met: first, it filters the set of allowed candidate control modes according to the task constraint policy information, then it performs a state feasibility determination on each candidate control mode according to the entry conditions in the mode determination rule base, and finally determines whether to switch the control mode or execute the preset backup action in the task constraint policy information when there is no feasible candidate control mode.

[0157] 6. Example Description

[0158] This embodiment illustrates that in the transparent transmission mode, the remote control center 109 directly implements fine-grained real-time control over the UAV 108. The UAV nest does not undertake task splitting, task feasibility judgment, or local master control closed-loop functions, but mainly participates in the control process as a communication transmission node, status aggregation node, and take-off and landing auxiliary node. It is suitable for task scenarios that require continuous real-time human participation and fine operation.

[0159] Example 2: Edge Autonomous Execution Example of Standardized Unmanned Inspection Tasks

[0160] 1. Task creation and constraint loading

[0161] This embodiment uses an unmanned standardized power line inspection task as an example. Before execution, the remote control center 109 has completed the inspection route arrangement, waypoint action configuration, shooting rule configuration, and return strategy configuration, and then distributed the corresponding task package to the device nest. The task package includes a task type identifier, a preset route file, target point information, shooting action parameters, map data reference information, no-fly zone constraints, return threshold, and anomaly handling strategies. During the task creation phase, the remote control center 109 determines that the task is an unmanned standardized power line inspection task based on the task requirements and generates a corresponding task type identifier. The remote control center 109 retrieves the corresponding task constraint strategy information from the task constraint strategy library based on the task type identifier. This strategy information limits the task to a pre-arranged task type that allows local-led continuous execution. The task constraint strategy information is distributed to the edge computing controller 102 along with the task package.

[0162] 2. Determining the initial control state

[0163] Based on the aforementioned task attributes, the edge computing controller 102 powers on the UAV and collects necessary state information before the task begins. The edge computing controller 102 treats the task start preparation as a mode determination trigger event. After state evaluation and mode determination, it determines that the initial control mode of the task is edge autonomous mode.

[0164] 3. Task execution process

[0165] After the mission begins, the edge computing controller 102 controls the nest execution unit 107 to complete the hatch opening, release mechanism action, and takeoff preparation actions, and controls the UAV 108 to take off and execute the inspection task according to the locally cached schedule. During mission execution, the edge computing controller 102 maintains the current mission execution state machine and autonomously advances the mission execution according to the locally cached route, waypoint actions, and mission rules; the state awareness module 104 continuously collects the UAV operating status, nest equipment status, edge resource status, and network link status, and sends the collection results to the edge computing controller 102 for local master control and anomaly handling during mission execution.

[0166] 4. High-level intervention input processing logic

[0167] In this mode, the remote control center 109 can still receive mission progress, image data, and telemetry data via the wide area link, and can send new higher-level commands to the nest according to the needs of the upper-level business, such as requesting additional shooting, adjusting the priority of a certain area, requesting early return, requesting to terminate the current mission, or requesting to skip a non-critical waypoint. However, since the current mission has formed a continuous execution closed loop on the nest side, and the mission execution state machine is maintained by the edge computing controller 102, such commands issued by the remote control center 109 are no longer automatically effective process control commands, but are processed as higher-level intervention inputs.

[0168] 5. Local task feasibility assessment logic

[0169] After receiving the higher-level intervention input, the edge computing controller 102 needs to make a judgment based on the local task state machine and local rule base. Specifically, the edge computing controller 102 can combine the current task closed-loop constraints, the current task stage, the task priority strategy, preset preemption rules, preset return-to-home rules, preset recovery rules, and local real-time status parameters to determine the acceptance and execution timing of the command. The local real-time status parameters include at least the current remaining battery power, UAV position, speed, attitude, obstacle avoidance status, pod actuator status, recovery feasibility status, and on-site environmental status.

[0170] For example, when the remote control center 109 issues a command to "go to area D to perform supplementary photography" during an inspection task, the edge computing controller 102 first determines, based on local rules, whether the command will disrupt the current main task loop, whether it meets the local priority preemption conditions, and whether it conforms to the preset return and recovery rules. If, based on the local real-time status, it is determined that although the command meets the execution conditions, it will interrupt the current critical inspection loop and its priority is insufficient to preempt the current main task, then the edge computing controller 102 may not immediately accept the command, but instead suspend, queue, or directly report the rejection reason. If it is determined that the command is a high-priority supplementary photography and the local real-time status meets the insertion execution conditions, then the edge computing controller 102 may insert it into the current execution process at an appropriate time.

[0171] 6. Example Description

[0172] This embodiment illustrates that in the scenario of unattended standardized inspection tasks, the core of the edge autonomous mode is that the task execution state machine is maintained by the nest side, and the main judgment of task feasibility is undertaken by the nest side. Although the remote control center 109 can still issue high-level intervention commands, whether the command enters the current task execution process is determined by the edge computing controller 102 after judging the task feasibility based on the local task status and task constraint strategy information.

[0173] Example 3: Example of manual inspection tasks executed in intelligent relay mode

[0174] 1. Task creation and constraint loading

[0175] This embodiment uses a manually-participated temporary inspection task as an example. Such tasks typically rely on the remote control center 109 to continuously receive on-site image transmissions, telemetry data, and status information. Remote operators dynamically generate the next task requirements based on task progress and on-site conditions. Therefore, during the task creation phase, the remote control center 109 determines the task to be a manually-participated temporary inspection task based on the task requirements and generates a corresponding task type identifier. The remote control center 109 retrieves the corresponding task constraint policy information from the task constraint policy library based on the task type identifier. This policy information restricts the task to continuous decision-making by the remote control center 109 and allows intelligent relay mode. The task constraint policy information is sent to the edge computing controller 102 along with the task package.

[0176] 2. Determining the initial control state

[0177] Based on the aforementioned task characteristics, the edge computing controller 102 powers on the UAV and collects necessary status information before the task begins. The edge computing controller 102 treats the task start preparation as a mode determination trigger event. After status evaluation and mode determination, it determines that the initial control mode of the task is intelligent relay mode.

[0178] 3. Task execution process

[0179] During mission execution, the remote control center 109 continuously receives status data from the nest and the UAV 108 via the multi-link communication module 103, including current position, flight attitude, remaining battery power, image transmission content, mission completion status, and nest status information. Based on this relatively complete real-time status, the remote control center 109 undertakes the primary judgment of mission feasibility and advanced mission decisions. In this mode, the mission state machine is maintained by the remote control center 109, and the primary judgment authority for mission feasibility belongs to the remote control center 109. For example, the remote control center 109 can determine the next target location, the next observation area, the next shooting action requirement, or the execution sequence of the next partial flight segment based on the on-site images, and generate corresponding advanced mission instructions.

[0180] 4. Control logic in intelligent relay mode

[0181] Upon receiving the high-level task instruction, the edge computing controller 102 does not primarily determine whether the task should be executed. Instead, it mainly converts the high-level task instruction into a locally executable control sequence. Specifically, the edge computing controller 102 can perform task decomposition, local path organization, protocol conversion, and flight control instruction decomposition on the high-level task instruction based on the current location of the UAV 108, the target area location, local map data, and local flight control interface requirements. It then sends corresponding control instructions to the UAV 108 via the local wireless link. Simultaneously, the edge computing controller 102 can also perform execution-level protective checks, such as checking the validity of the instruction format, the availability of the local interface, the normality of the flight control link, and the basic executability of the generated local path.

[0182] For example, after the remote control center 109 issues an advanced mission command to "fly to area A and complete three fixed-point photography", the edge computing controller 102 can convert the command into a local flight path to the target area, three photography action points, and the corresponding underlying flight control command sequence, and organize the UAV 108 to execute it. The telemetry data and payload data generated by the UAV 108 during execution are aggregated through the drone nest and transmitted back to the remote control center 109 to support the remote control center 109 in continuing to make the next task judgment.

[0183] In this mode, the mission execution state machine is maintained by the remote control center 109, and the nest primarily undertakes functions such as high-level mission instruction translation, local planning, flight control instruction decomposition, and execution-level protective verification. If the execution-level protective verification detects obvious anomalies, such as the local interface being unavailable, local path generation failing, or flight control channel malfunctioning, the edge computing controller 102 can temporarily suspend execution and report the abnormal status to the remote control center 109; however, this type of processing belongs to execution-level protection and does not change the control relationship in the intelligent relay mode where the remote control center 109 is responsible for the primary judgment of mission feasibility.

[0184] 5. Example Description

[0185] This embodiment illustrates that in the intelligent relay mode of manual inspection tasks, the remote control center 109 is responsible for making the main judgment on task feasibility and task decision based on a relatively complete real-time status. The nest side does not undertake the main judgment at the task level, but mainly undertakes the functions of high-level instruction translation, local execution organization, and execution-level protective verification.

[0186] Example 4: Example of switching from intelligent relay mode to edge autonomy mode

[0187] 1. Task creation and task constraint policy information distribution

[0188] This embodiment uses the scenario of gradual degradation of wide area links during the execution of manual inspection tasks as an example for illustration. Such tasks typically rely on the remote control center 109 to continuously receive on-site image transmission, telemetry data, and task status information, and the remote operator generates and issues the next business command sequentially based on the results of the previous task execution, the current image content, and changes in the on-site situation.

[0189] During the task creation phase, the remote control center 109 determines that the task is a manual inspection task based on the task requirements and generates a corresponding task type identifier. Based on the task type identifier, the remote control center 109 retrieves the task constraint policy information corresponding to the manual inspection task from the task constraint policy library and sends the task type identifier and task constraint policy information to the data center side along with the task package.

[0190] The task constraint policy information includes at least the set of control modes that can be switched, the set of control modes that cannot be switched, whether remote continuous decision-making is required, whether local takeover is allowed, whether continued execution is allowed after link failure, the authority to transfer the main judgment of task feasibility, the flight phases that can be switched, the flight phases that cannot be switched, the priority of the target control mode, and the backup actions when the link deteriorates or there is no feasible mode.

[0191] 2. Determining the Initial Control Mode of the Task

[0192] During the mission initiation preparation phase, the nest-side edge computing controller 102 powers on the UAV 108 and collects necessary status information before mission initiation. This necessary status information includes network link status, edge resource status, UAV initial status, nest energy status, nest actuator status, local mission template integrity, local map data integrity, and local mission context integrity.

[0193] The edge computing controller 102 treats task startup preparation as a mode determination trigger event, and performs initial control mode determination based on task constraint policy information, mode determination rule base and necessary state information before task startup.

[0194] In this embodiment, the edge computing controller 102 detects that the wide-area link quality meets the requirements of the remote control center 109 to continuously receive status information, enabling the remote control center 109 to obtain relatively complete image transmission, telemetry data, and task status information. Simultaneously, the task constraint policy information allows the remote control center 109 to continuously undertake the primary judgment of task feasibility and high-level task decision-making during manual inspection tasks. Based on this, the edge computing controller 102 determines the intelligent relay mode from the candidate control modes as the initial control mode for the formal execution of the task.

[0195] 3. Task execution in intelligent relay mode

[0196] After the mission begins, the system enters intelligent relay mode. In this mode, the remote control center 109 acts as the primary control entity, responsible for maintaining the mission state machine and possessing the primary authority to determine mission feasibility. The remote control center 109 continuously receives real-time status parameters from the nest side and the UAV 108 through the multi-link communication module 103, including current position, flight attitude, remaining battery power, mission completion status, image transmission content, nest status information, and link status information, and undertakes the primary judgment of mission feasibility and high-level mission decision-making based on these status parameters.

[0197] After receiving a service command from the remote control center 109, the edge computing controller 102 does not undertake the main judgment at the task layer, but is mainly responsible for converting the service command into a locally executable control sequence. Specifically, the edge computing controller 102 can perform task decomposition, local path organization, protocol conversion, and flight control instruction decomposition of the service command according to the current location of the UAV 108, the target area location, map data, and flight control interface requirements, and then send corresponding control commands to the UAV 108 through the local wireless link.

[0198] Meanwhile, the edge computing controller 102 performs execution-level protective checks, such as checking whether the command format is valid, whether the local interface is available, whether the flight control channel is normal, and whether the local path has basic executability.

[0199] In this type of manual inspection task, the remote control center 109 typically does not issue multiple consecutive subsequent tasks at once. Instead, it generates and issues the next business command sequentially based on the execution result of the previous task, the transmitted images, the task completion report, and the on-site situation. For example, after the drone 108 completes the shooting at the current observation point and transmits the completion status back to the remote control center 109, the remote control center 109 generates a new business command based on the received information and issues it to the drone nest side. The edge computing controller 102 performs local path organization, flight control command decomposition, and execution organization for the business command.

[0200] 4. Pattern Determination Triggering and State Evaluation

[0201] During mission execution, the status awareness module 104 continuously collects network link status, mission status, edge resource status, and UAV operating status.

[0202] At a certain moment, the state awareness module 104 detects a continuous increase in latency, packet loss rate, and link jitter in the wide-area link, causing the on-site status received by the remote control center 109 to gradually become lagging, incomplete, or discontinuous. Based on this, the edge computing controller 102 determines that the network link status change meets the pattern determination triggering condition and initiates the state assessment and determination input construction process.

[0203] The edge computing controller 102 constructs a set of input parameters for pattern determination based on the latest collected multi-source state information. This set of input parameters includes at least the network quality comprehensive index Qnet, the duration of continuous degradation of the wide-area link, the integrity of the status reception of the remote control center 109, the status of local edge resources, the integrity of the local task template, the integrity of the local map data, the integrity of the local task context, the validity of local control rules, the remaining battery power of the UAV 108, the flight status of the UAV 108, the current task stage, the task progress, and the nest recovery capability.

[0204] In this embodiment, the edge computing controller 102 detects that: the target information, action requirements, local path parameters, map data, return parameters, safety constraints and anomaly handling strategies required for the currently executed task have been fully obtained and cached in the data storage unit 105; the local edge resources are in normal condition; and the current remaining battery power, flight status and nest recovery capability of the drone 108 meet the conditions for completing the current task or advancing the current task to the preset safe end node.

[0205] 5. Candidate control mode screening and target control mode selection

[0206] After forming the set of input parameters for pattern determination, the edge computing controller 102 first determines the set of candidate control modes allowed for the current task based on the task constraint policy information.

[0207] In this embodiment, the task constraint policy information allows the nested side to primarily determine task feasibility when the wide-area link is significantly degraded and the local state meets the autonomous execution requirements. Therefore, the edge computing controller 102 includes the edge autonomous mode in the candidate control mode set.

[0208] Subsequently, the edge computing controller 102 invokes the mode determination rule base to determine the feasibility of each candidate control mode. Since the wide area link has not been completely interrupted, but the network quality index Qnet is lower than the maintenance requirement of the intelligent relay mode, and the edge resource status, local task template, local map data, local task context, and UAV operating status meet the entry conditions for the edge autonomous mode, the edge computing controller 102 determines that the edge autonomous mode meets the entry conditions under the current state.

[0209] When the link degradation state continues for a preset hysteresis time, the edge computing controller 102 determines that the target control mode is edge autonomous mode. Since the target control mode is inconsistent with the current intelligent relay mode, the system enters the control mode switching execution and state synchronization process.

[0210] 6. Control mode switching execution and status synchronization

[0211] When switching from intelligent relay mode to edge autonomous mode, the edge computing controller 102 performs control handover, state synchronization and instruction semantic level switching process.

[0212] First, the edge computing controller 102 performs pre-switch verification on the current task stage, the UAV's safety status, the conditions for entering edge autonomous mode, and the main control capabilities on the nest side. Once the pre-switch verification passes, the edge computing controller 102 freezes the current queue of instructions to be executed, suspends the activation of new process control instructions, and generates a task status snapshot.

[0213] The mission status snapshot includes the current control mode, target control mode, current main control entity, target main control entity, mission state machine status, mission progress, current mission breakpoint, current waypoint or work point, UAV 108 position, speed, attitude, battery level, payload status, alarm status, key control parameters, current mission feasibility assessment authority, and current command semantic level.

[0214] Then, the edge computing controller 102 prioritizes synchronizing alarm information, key telemetry information, task progress information, and task completion status, and performs a consistency comparison between the task status recorded locally in the drone nest and the task status recorded by the remote control center 109. This consistency comparison is used to ensure that the remote control center 109 and the drone nest have a consistent understanding of the task progress, task breakpoints, key drone status, and command execution status, but does not change the drone nest's primary judgment authority on task feasibility in edge autonomous mode.

[0215] Once the consistency comparison passes, the edge computing controller 102 switches the task state machine maintenance entity from the remote control center 109 to the nest side, transfers the main judgment authority for task feasibility from the remote control center 109 to the edge computing controller 102, and updates the main control entity identifier, control command issuance channel, command parsing logic, and command semantic level.

[0216] At the same time, the system updates the semantic level of control commands that the remote control center 109 is allowed to use in the current control mode, so that the business commands issued by it can remain consistent in terms of surface task description, but the parsing semantics, processing flow and effective method on the nest side are switched from process control command mode to high-level target input, constraint input, priority adjustment input or intervention input mode.

[0217] In intelligent relay mode, the service command is a process control command after the remote control center 109 completes the main feasibility judgment of the task. The edge computing controller 102 mainly performs task splitting, local path organization, protocol conversion, flight control command decomposition, and execution-level protective verification. In edge autonomous mode, the service command must first enter the local acceptance judgment and execution timing judgment process, and then the edge computing controller 102 determines whether to include it in the execution process based on the local real-time status.

[0218] Subsequently, after confirming that the state takeover has been completed on the nest side, the edge computing controller 102 resumes the instruction execution queue and continues to execute tasks according to the control rules corresponding to the edge autonomous mode.

[0219] 7. Execution logic after switching

[0220] After the handover is complete, the remote control center 109 can still make a pre-judgment of whether to send the next business command based on the task completion report it has received, historical telemetry information, and existing task rules. For example, the remote control center 109 can continue to send the business command "go to area C to take a picture".

[0221] However, in edge autonomous mode, the service command no longer represents a process control instruction that takes effect automatically, but rather a high-level target request or intervention input from the remote control center 109 to the nested side. After receiving the service command, the edge computing controller 102 no longer directly organizes execution based solely on the result sent by the remote control center 109, but instead combines local real-time status parameters to determine the feasibility of the task.

[0222] The local real-time status parameters include at least the current remaining battery power, UAV location, speed, attitude, obstacle avoidance status, payload status, nest actuator status, recovery feasibility status, and on-site environmental status. When necessary, the edge computing controller 102 can also adjust the execution timing of the service command based on the current local task state machine, deciding whether to execute immediately, postpone execution, suspend execution, refuse execution, or initiate a return-to-home process. That is, the same service description "go to area C to take photos" represents a process control command after the remote control center 109 has completed the primary feasibility assessment in intelligent relay mode; while in edge autonomous mode, it only represents a high-level target input that needs to be judged by the nest side in conjunction with the local status. The same service command corresponds to different semantic levels in different control modes, and whether and when it is ultimately executed is determined by the edge computing controller 102 in conjunction with the local status.

[0223] For example, after the remote control center 109 sends a service command to "go to area C to take pictures" based on the received previous task completion report, the edge computing controller 102 may not immediately execute the service command. Instead, it may send a message to the remote control center 109 indicating that although the remaining power is close to the threshold, it is insufficient to cover the new target area for taking pictures and then safely return to base. Alternatively, if the current status of the nest recovery mechanism does not meet the subsequent recovery requirements, or if the on-site environment is no longer suitable for continuing the service command, the edge computing controller 102 may not immediately execute the service command. Instead, it may send a message to the remote control center 109 indicating whether the execution is suspended, delayed, switched to base, or refused to execute.

[0224] If the edge computing controller 102 determines that the current local real-time status meets the execution conditions, it can incorporate the business command into the current local task execution process and organize the drone 108 to execute it.

[0225] 8. Example Description

[0226] This embodiment illustrates that the essence of switching from intelligent relay mode to edge autonomous mode is as follows: In a manual inspection scenario where tasks are issued one by one, when the degradation of the wide area link causes the real-time parameters received by the remote control center 109 to be no longer sufficiently reliable, the system no longer relies on the remote control center 109 to make the primary judgment on task feasibility. Instead, the edge computing controller 102 takes over the primary judgment of task feasibility and, in conjunction with the local real-time status on the nest side, makes judgments on the acceptance of subsequent business commands, the timing of execution, and security.

[0227] Thus, the system has switched from "remote control center making decisions and the machine nest translating and executing" to "remote control center making predictions and the machine nest making decisions", which enables the continuity, security and consistency of task execution and control logic to be maintained even when the wide area link degrades but is not completely interrupted.

[0228] Example 5: Example of switching from edge autonomous mode to independent master control mode

[0229] 1. Task creation and task constraint policy information distribution

[0230] This embodiment uses a scenario where a wide-area link is continuously interrupted during the execution of a standardized inspection task as an example. At the start of the task, the edge computing controller 102 determines, based on the task type identifier, that the task is a pre-arranged task that is allowed to be executed continuously locally.

[0231] During the task creation phase, the remote control center 109 generates a task type identifier based on the task type and retrieves the corresponding task constraint policy information from the task constraint policy library. The task type identifier and task constraint policy information are sent to the nest side along with the task package. The task constraint policy information includes at least the set of allowed control modes, the set of prohibited control modes, whether continued execution is allowed after a link failure, whether local takeover is allowed, the allowed instruction semantic level during mode switching, the authority to transfer the primary judgment of task feasibility, the flight phases during which switching is allowed and prohibited, the priority of the target control mode, and the backup actions in case of link degradation or no feasible mode.

[0232] 2. Determining the Initial Control Mode of the Task

[0233] During the mission startup preparation phase, the nest-side edge computing controller 102 controls the UAV 108 to power on and collects necessary status information before startup, including network link status, edge resource status, UAV startup status, nest energy status, nest actuator status, local mission template integrity, local map data integrity, and local mission context integrity.

[0234] The nest side treats task initiation preparation as a mode determination trigger event, and combines task constraint policy information, mode determination rule base, and current state to determine the initial control mode. In this embodiment, the initial mode determination result is edge autonomous mode, because the task type and task constraint policy allow the nest side to maintain the task state machine and undertake the main judgment of task feasibility without human intervention.

[0235] 3. Task execution in edge autonomous mode

[0236] In edge-autonomous mode, the mission execution state machine and mission feasibility master judgment are maintained by the nest-side edge computing controller 102. The remote control center 109 can receive mission progress and status information via a wide area link and send higher-level intervention inputs when necessary. The edge computing controller 102 combines local real-time status parameters to perform command parsing, mission decomposition, local path organization, flight control command decomposition, and execution-level protective verification.

[0237] 4. Pattern Determination Triggering and State Evaluation

[0238] During task execution, the state awareness module 104 continuously collects network link status, task status, edge resource status, and UAV operating status. When the wide-area link is continuously interrupted and the interruption duration exceeds a preset threshold, the edge computing controller 102 determines that the mode determination trigger condition is met and initiates the state assessment and determination input construction process.

[0239] The constructed set of input parameters for pattern determination includes the duration of continuous network link interruption, the integrity of remote control center status reception, the status of local edge resources, the integrity of local task templates, the integrity of local map data, the integrity of local task context, the remaining battery power of UAV 108, the flight status of UAV 108, the current task stage, the task progress, and the nest recovery capability.

[0240] 5. Candidate control mode screening and target control mode selection

[0241] The edge computing controller 102 determines a set of candidate control modes based on task constraint policy information. Since the wide-area link has been continuously interrupted and the task constraint policy allows execution to continue after the link is broken, the independent master control mode is included in the set of candidate control modes.

[0242] The edge computing controller 102 performs state feasibility determination on each candidate control mode based on the mode determination rule base. The edge autonomous mode is no longer suitable for maintaining the current task execution, and the independent master control mode meets the entry conditions. Therefore, the target control mode is determined to be the independent master control mode.

[0243] 6. Control mode switching execution and state synchronization

[0244] The target control mode differs from the current edge autonomous mode. The system enters a process of control handover, state synchronization, and instruction semantic level switching. The edge computing controller 102 freezes the current queue of instructions to be executed, records the current task breakpoint, flight status, remaining waypoints, battery status, and payload status; disables the dependency on the real-time task session of the remote control center 109; and updates the current task main control mode identifier.

[0245] Subsequently, the nest side reconstructs the subsequent task execution queue based on the local task template and cached task fragments, and synchronously updates the semantic level of control instructions to ensure that the business commands issued remotely are consistent in the surface task description, but are parsed as high-level target inputs or intervention inputs on the nest side, and the nest determines whether to execute them based on the local state.

[0246] 7. Task execution in independent master control mode

[0247] After entering independent master control mode, the edge computing controller 102 independently completes subsequent tasks based entirely on local control rules. If the remaining power is sufficient to cover all remaining waypoints, it continues to complete the inspection points; if the power is insufficient, it prioritizes completing critical waypoints and triggers the return to base; if the conditions for homing recovery change or the environment becomes abnormal, it aborts the remaining tasks and prioritizes ensuring a safe return to base and recovery.

[0248] After the UAV 108 returns to base, the nest execution unit 107 completes the hatch opening, recovery docking, centering reset, and power replenishment actions. The mission logs, payload data, and mission status generated during the link loss are cached in the data storage unit 105. During this stage, the mission execution state machine and the main judgment of mission feasibility are maintained independently by the nest side, without the need for the participation of the remote control center 109.

[0249] 8. Determining and Selecting the Mode for Network Recovery Trigger

[0250] When the wide-area link recovers and stabilizes to a preset threshold, the state awareness module 104 detects that the network recovery meets the mode determination trigger condition, and the edge computing controller 102 initiates the mode determination process. Based on the latest collected multi-source state information, including network link status, edge resource status, UAV operating status, and task status, combined with task constraint policy information and the mode determination rule base, the system filters the candidate control modes allowed for the current task and determines the target control mode.

[0251] After the determination is completed, if the target control mode is consistent with the current independent master control mode, the current control mode will continue to be maintained; if the target control mode is different from the current control mode, the control handover, state synchronization and instruction semantic level switching process will be executed to switch the system to the new mode; if there is no candidate control mode that meets the conditions, the preset backup action in the task constraint policy information will be executed.

[0252] Through this mechanism, the system can automatically select the most suitable control mode for the current task and state after the wide area link is restored, so as to realize dynamic adjustment of control mode and ensure task continuity, while ensuring the consistency of control authority, task state machine and instruction semantics.

[0253] Furthermore, in the above embodiments, the calculation method of the comprehensive network quality index, the entry and exit thresholds for each mode, the hysteresis time length, the disconnection threshold, the recovery threshold, the edge resource occupancy threshold, the return-to-home battery power threshold, and the definition of switchable flight phases can all be adjusted according to the actual application scenario, aircraft capabilities, mission risk level, and deployment environment. Without changing the technical concept of this invention—"adaptive switching of multi-level control modes based on a task constraint strategy library, control handover, and synchronous switching of instruction set levels"—any adjustments to the parameter format, module implementation format, or process sequence should be considered equivalent implementations of this invention.

Claims

1. A method for adaptive switching of multi-level control modes based on task constraints, characterized in that, Includes the following steps: Predefine multi-level control modes and configure task constraint policy library, mode determination rule library and mode determination trigger conditions; The remote control center generates a task type identifier based on the task type, retrieves the corresponding task constraint policy information from the task constraint policy library, and sends it to the nest side; the nest side edge computing controller collects necessary status information during the task startup preparation phase and determines the initial control mode. During mission execution, the nest continuously collects multi-source status information; The edge computing controller determines whether the mode determination triggering condition is met based on the collected multi-source state information; if the condition is met, it initiates state evaluation and constructs a set of mode determination input parameters; if the condition is not met, it maintains the current control mode. The edge computing controller determines a set of candidate control modes based on task constraint policy information, and performs state feasibility determination on the candidate control modes based on the set of mode determination input parameters and the mode determination rule base to determine the target control mode. If the target control mode is different from the current control mode, then control handover, state synchronization, and instruction semantic level switching are performed; if there is no candidate control mode that meets the conditions, then a backup action is performed. Before the edge computing controller performs a mode switch check, it freezes the current instruction queue and generates a task state snapshot, and updates the main control entity identifier, the task state machine maintenance entity, the main entity with the main judgment authority for task feasibility, the control instruction issuance channel, the instruction parsing logic, and the instruction semantic level; after the mode switch is completed, it restores the instruction execution queue.

2. The method of claim 1, wherein, The multi-level control modes include transparent transmission mode, intelligent relay mode, edge autonomous mode and independent master control mode. Each mode is distinguished based on the task state machine maintaining entity, the ownership of the main judgment authority for task feasibility and the semantic level of control instructions. The task constraint policy library stores task constraint policy information corresponding to each task type, including the set of control modes that can be switched, the priority of the target control mode, and the backup action; the mode determination rule library is configured on the nest side and defines the entry conditions, maintenance conditions, and exit conditions of each control mode; the multi-source status information includes network link status information, task status information, edge resource status information, and UAV operation status information.

3. The method of claim 2, wherein, In the transparent transmission mode, the remote control center acts as the main control entity, directly issuing low-level flight control commands to the UAV; in the intelligent relay mode, the remote control center acts as the main control entity, and the UAV's nest splits, translates, and performs protective verification on the task commands issued by the remote control center according to their type; in the edge autonomy mode, the UAV's nest acts as the main control entity, and the commands from the remote control center serve as target inputs or intervention inputs; in the independent master control mode, the UAV's nest independently controls itself based on local data when the wide-area link cannot meet the remote control requirements.

4. The method of claim 1, wherein, The triggering conditions for mode determination include: network link triggering conditions, edge resource triggering conditions, task phase triggering conditions, task constraint triggering conditions, UAV operating status triggering conditions, and periodic scheduling triggering conditions.

5. The method of claim 2, wherein, The set of input parameters for pattern determination includes multi-source state information and the results of state assessment. The state assessment includes network link state assessment, edge resource availability assessment, UAV operation state assessment, and mission state assessment.

6. The method of claim 5, wherein, The network link state evaluation includes constructing a network quality comprehensive index for representing a degree of support of the current link to the remote control capability. ； wherein, is a delay normalization function, is a packet loss rate, is a bandwidth normalization function, is a jitter normalization function, is a weight coefficient; The edge resource availability assessment includes evaluating processing capacity, storage capacity, local task template integrity, local map data integrity, local task context integrity, local control rule validity, nest actuator status, and nest energy status; used to determine whether the nest side has autonomous execution or independent control capabilities. The drone operation status assessment includes assessing battery status, position, speed, attitude, payload status, and obstacle avoidance status, in order to determine whether the drone itself is in a safe state suitable for switching. The task status assessment includes evaluating the current task stage, task progress, whether remote continuous decision-making is still required, and whether the conditions for offline continuation are met, which provides task-side status input for subsequent candidate control mode screening.

7. The method of adaptive switching between multi-level control modes based on task constraints according to claim 6, characterized in that, The rules for determining the feasibility of the state include: Conditions for entering transparent pass-through mode: ≥T1, and the remote control center can continuously obtain complete status information; T1 is the link quality threshold that supports transparent pass-through mode; Entry conditions for the intelligent relay mode: T2 ≤ <T1, the remote control center can continuously obtain complete status information, and the nest has the capabilities of task splitting, protocol conversion, and instruction translation; T2 is the link quality threshold for supporting the intelligent relay mode; Entry condition of edge autonomous mode: T2 and wide area link has not been completely interrupted, edge resource status, local task template, local map data and task context meet autonomous execution condition; Entry conditions for independent master control mode: The duration of continuous interruption of wide area link exceeds the preset threshold T3, and the edge resource status, local task template, task context, local control rules and drone operation status meet the independent execution conditions; When multiple candidate control modes meet the feasibility conditions, the system determines the most suitable target control mode according to the priority of the target control mode, the current task stage, the task continuity requirements, and the security requirements in the task constraint strategy information. When the target control mode is consistent with the current control mode, the system maintains the current control mode and continues to execute tasks and monitor the status. When the target control mode is inconsistent with the current control mode, the system enters the control mode switching process, namely, the handover of control, status synchronization, and instruction semantic level switching. When the target control mode switching conditions are detected, the current status is required to continue for a preset hysteresis time before the control mode switching and status synchronization are executed. When there is no candidate control mode that meets the conditions, or when a safety event such as low battery, obstacle avoidance trigger, abnormal positioning, or abnormal attitude occurs, causing the current task to be unable to continue to be executed according to any candidate control mode, the system executes the preset backup action in the task constraint strategy information. The backup action includes one or more of the following: hovering and waiting, returning to home, landing, continuing the current safety action, waiting for manual takeover, or terminating the task.

8. The method of claim 2, wherein, During mode switching, for cases where the target control mode is transparent transmission mode, intelligent relay mode, or edge autonomous mode, a consistency comparison is performed between the local task status of the data center and the task status recorded by the remote control center; for cases where the target control mode is independent master control mode, the consistency comparison is skipped, and the switching event and task progress are synchronized through background retransmission after the link is restored.

9. A mission-constrained multi-level control mode adaptive switching system for dynamic switching of control modes between a UAV nest and a remote control center using the method of any one of claims 1-8, characterized in that, include: Edge computing controller, multi-link communication module, status awareness module, data storage unit, energy management unit, nested execution unit, drone, and remote control center; The edge computing controller is the core control unit on the nest side, which realizes adaptive switching of control mode and dynamic handover of control, and parses and executes control commands according to the current control mode. The multi-link communication module is used to establish a local wireless communication link between the nest and the UAV, as well as a wide-area communication link between the nest and the remote control center, and to transmit control commands, telemetry data, mission status data and payload data. The state awareness module is used to collect multi-source state information related to task execution and send it to the edge computing controller; the multi-source state information includes network link status, nest device status, edge resource status, and UAV operating status; The data storage unit is used to store task templates, flight path files, map data, no-fly zone data, task breakpoint information, recovery parameters, and telemetry data, video data, and log data generated during task execution, in order to support task execution, mode switching, and disconnection recovery control. The energy management unit is used to manage the power supply status of the nest and the UAV, and to provide power supply status, charging status and battery status information to the edge computing controller for mission take-off determination, endurance assessment and return triggering. The nest execution unit is used to execute nest mechanism actions under the control of the edge computing controller to cooperate with the UAV to complete the take-off, landing, recovery, reset and recharging processes; The remote control center is used to create tasks, generate task packages, issue control commands, and remotely supervise and manage the task execution process. The remote control center is used to generate task type identifiers, task content, and task constraint information, and distribute them to the drone nest side via a wide area network. At the same time, it receives telemetry data, payload data, and task status information transmitted back from the drone nest side and the drone to support task monitoring and anomaly handling.

10. The task constraint based multi-level control mode adaptive switching system according to claim 9, wherein, The edge computing controller is connected to the multi-link communication module, the state awareness module, the data storage unit, the energy management unit, and the nest execution unit, respectively. The multi-link communication module is connected to the UAV and the remote control center to realize local wireless communication and wide area network communication.

Images