Autonomous decision optimization method for aerial robot

Abstract

The invention relates to the technical field of robot navigation, in particular to an autonomous decision optimization method for an aerial robot. Acquiring flight environment data, calculating an obstacle density factor, a weather influence factor and a height risk factor, and generating an environment influence parameter for dynamically adjusting a control strategy; constructing a hybrid control framework which comprises two parallel branches of global path tracking control and local obstacle avoidance control, and dynamically adjusting a control weight according to environmental influence parameters; a three-layer hierarchical control architecture is established, the three-layer hierarchical control architecture comprises a macroscopic task constraint layer, a mesoscopic maneuvering decision layer and a microscopic flight control layer, upper layer constraints are converted into bottom layer control instructions step by step by controlling a constraint propagation mechanism, and consistency and safety verification is carried out; and a state-weight mapping controller is adopted to dynamically adjust a multi-target optimization weight coefficient according to the real-time state vector, and an optimal control instruction is generated. According to the invention, the autonomous decision-making capability and the flight safety of the aerial robot in a dynamic environment are improved.

Description

Technical Field

[0001] This invention relates to the field of robot navigation technology, specifically to an autonomous decision-making optimization method for aerial robots. Background Technology

[0002] Existing decision-making optimization methods for aerial robots mainly suffer from the following technical shortcomings: Traditional deep neural networks use fixed feature extraction layers, which cannot adaptively adjust the granularity and focus of feature representations according to changes in the complexity of the aerial environment. When faced with different flight altitudes, weather conditions, and obstacle densities, the fixed feature extraction method leads to the loss of key information or overload of redundant information, affecting the accuracy of decision-making.

[0003] Existing knowledge reasoning techniques discretize continuous three-dimensional space into a fixed grid for symbolic representation. This coarse-grained discretization method is difficult to accurately express the continuity constraints of the flight path of aerial robots and the real-time changes of dynamic obstacles, resulting in insufficient path planning and inadequate safety assessment.

[0004] Existing methods employ a pre-defined fixed weight allocation strategy when dealing with multiple decision objectives such as safety, efficiency, and energy consumption. This fails to dynamically adjust decision preferences based on real-time flight status, mission urgency, and environmental risks, resulting in inflexible and poorly adaptable decision outcomes.

[0005] To address this, an autonomous decision-making optimization method for aerial robots is proposed. Summary of the Invention

[0006] The purpose of this invention is to provide an autonomous decision-making optimization method for aerial robots.

[0007] To achieve the above objectives, the present invention provides the following technical solution: An autonomous decision-making optimization method for aerial robots includes: The system acquires flight environment data of aerial robots and generates environmental impact parameters for dynamically adjusting control strategies by calculating obstacle density factors, weather impact factors, and altitude risk factors. Based on the environmental impact parameters, in a hybrid control framework that includes global path tracking control and local obstacle avoidance control, the control weights of the two are dynamically adjusted through parallel control branches. In low-impact environments, priority is given to ensuring the optimality of the global trajectory, while in high-impact environments, priority is given to ensuring the safety of local avoidance, thereby achieving adaptive allocation of flight control focus. In a hierarchical control architecture consisting of a macro-level mission constraint layer, a meso-level maneuver decision layer, and a micro-level flight control layer, the flight constraints and permitted maneuver behaviors defined at the upper level are transformed into specific aircraft control commands at the lower level through control constraint propagation, and the consistency and safety of the control commands are verified. Based on the real-time state vector of the aircraft, the weight coefficients of multiple optimization objectives in the flight control law are dynamically adjusted by the state-weight mapping controller to generate the optimal control command and apply it to the aircraft.

[0008] Furthermore, the flight environment data includes obstacle data, meteorological data, and terrain and airspace data.

[0009] Furthermore, the process of generating environmental impact parameters for dynamically adjusting control strategies by calculating obstacle density factors, weather impact factors, and altitude risk factors is as follows: The degree of interference of obstacles on trajectory planning is assessed by using a three-dimensional spatial occupancy grid map, and the obstacle density factor is calculated. The impact of wind speed, visibility, and precipitation intensity information on flight stability is assessed, and weather impact factors are calculated. Assess the risks of altitude selection by combining airspace restrictions and terrain undulations, and calculate the altitude risk factor; Finally, these assessment results were integrated into quantitative environmental impact parameters according to the control strategy of prioritizing obstacle density, secondary weather impact, and supplementing with high risk.

[0010] Furthermore, the process of adaptively allocating flight control focus through parallel control branches in the hybrid control framework is as follows: a small-range perception control branch uses a fine receptive field to extract control information for real-time obstacle avoidance, and a large-range perception control branch uses a wide-area receptive field to extract navigation control information for global path tracking; the environmental impact parameter is used to adjust the contribution of each control branch's information to the final control decision. When the environmental impact parameter is low, the weight of the large-range control branch is increased to focus on global navigation control, and when the environmental impact parameter is high, the weight of the small-range control branch is increased to focus on local obstacle avoidance control.

[0011] Furthermore, the hierarchical control architecture, consisting of a macro-level mission constraint layer, a meso-level maneuver decision-making layer, and a micro-level flight control layer, specifically includes: The macro-level mission constraint layer defines the geometric boundaries, no-fly zones, and airspace control rules for flight missions, and uses a geometric-semantic hybrid representation to store airspace control constraints. The meso-level maneuver decision layer provides a set of permissible flight maneuver control actions based on macro-level control constraints, and uses fuzzy production rules to represent maneuver control decisions. The micro-level flight control layer accurately converts selected maneuvers into direct control parameters for the aircraft.

[0012] Furthermore, the process of propagating control constraints step by step into specific aircraft control commands at the lower levels, and verifying the consistency and security of these control commands, is as follows: The control constraints of the macro-level task constraint layer are used to filter the maneuver control action options in the meso-level maneuver decision layer, calculate the degree of constraint violation for each control action, and generate control action weights. The micro-level flight control layer maps the maneuver control actions selected by the meso-level maneuver decision layer to a set of initial control parameters based on the current flight control state. Before outputting to the actuator, the control parameters are checked for safety boundaries to ensure that they will not cause the aircraft to stall or exceed the attitude control limits, and safety control corrections are made.

[0013] Furthermore, the process of dynamically adjusting the weight coefficients of multiple optimization objectives in the flight control law through a state-weight mapping controller, based on the real-time state vector of the aircraft, includes: Construct a real-time control state vector that includes the aircraft's own energy state, kinematic state, and mission and environment-related states; where the energy state includes battery percentage, the kinematic state includes flight altitude and current speed, and the mission environment state includes mission urgency score, weather risk level, and obstacle density. The state-weight mapping controller employs a control state encoder and a control weight generator to convert the control state vector into a four-dimensional control weight output that considers safety, efficiency, energy consumption, and accuracy. Specifically, the control state encoder compresses the six-dimensional control state vector sequentially through the first, second, and third control processing layers to convert it into a control state feature representation. The control weight generator then converts it into a four-dimensional control weight output through the fourth control processing layer and the output control processing layer.

[0014] Furthermore, the dynamic adjustment mechanism of the state-weight mapping controller specifically includes: The state-weight mapping controller takes the real-time control state vector of the aircraft as input and outputs a set of weight coefficients to balance the four control objectives of safety, efficiency, energy consumption and accuracy. The controller has a built-in safety priority control mechanism. When a high-risk control state is detected, such as low battery and being far from the landing point or an impending collision, the weight of the safety control objective is forcibly increased. At the same time, a smooth switching mechanism for control focus is established, which smooths the control weights through a time sliding window.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By using the parallel control branches in the hybrid control framework, local obstacle avoidance and global navigation information are extracted by small-range and large-range perception control branches respectively. The weight contribution of each branch is dynamically adjusted according to the environmental impact parameters, which enables adaptive allocation of flight control focus. This allows for optimization of global path tracking in low-risk environments and priority to ensure local obstacle avoidance safety in high-risk environments, thereby improving the flexibility and robustness of the overall flight control of aerial robots.

[0016] 2. By constructing a hierarchical control architecture consisting of a macro-level task constraint layer, a meso-level maneuver decision layer, and a micro-level flight control layer, which handles task constraints, maneuver decisions, and control parameter generation respectively, complex flight missions can be decomposed into clear control layers. This enables the orderly transmission of constraints from higher levels to lower levels of control, thereby improving the efficiency of control command generation and the structured and controllable nature of decisions. Through the propagation of control constraints into lower-level control commands and the verification of consistency and safety, it is possible to ensure that higher-level constraints effectively constrain lower-level decisions. At the same time, safety boundary checks and corrections prevent aircraft stall or attitude loss, thereby improving the reliability of control commands and the safety of the flight process.

[0017] 3. By constructing a real-time state vector that includes energy, kinematics, and mission environment states, and by using a state-weighted mapping controller to dynamically adjust the weight coefficients of safety, efficiency, energy consumption, and accuracy, the control law can be optimized according to the real-time state of the aircraft, so that the control commands can better balance the needs of multiple objectives, thereby improving the adaptability of decision-making and the quality of flight mission completion. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an autonomous decision-making optimization method for an aerial robot according to the present invention. Figure 2 This is a flowchart illustrating the obstacle density factor calculation process of the present invention; Figure 3 This is a schematic diagram of the state-weight mapping controller of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figures 1 to 3 This invention provides an autonomous decision-making optimization method for aerial robots, the technical solution of which is as follows: Example 1: To improve the intelligence of its aerial robots, a company used an autonomous decision-making optimization method for aerial robots proposed in this invention. The process of this method can be found by referring to [the following text is missing]. Figure 1 Specifically, it includes: The system acquires flight environment data of aerial robots and generates environmental impact parameters for dynamically adjusting control strategies by calculating obstacle density factors, weather impact factors, and altitude risk factors. Furthermore, flight environment data includes: obstacle data, meteorological data, and terrain and airspace data; Furthermore, the aerial robot acquires three-dimensional obstacle point cloud data, meteorological data, and ground elevation data through its lidar sensor, meteorological sensor, and RTK-GPS / INS integrated navigation system. The meteorological data includes wind speed, visibility, and precipitation intensity information. The terrain and airspace data includes not only ground elevation data but also airspace classification rules and airspace control information. By clearly defining flight environment data, including obstacle data, meteorological data, and terrain and airspace data, comprehensive environmental information input is provided for aerial robots, ensuring that the calculation of subsequent environmental impact parameters can accurately reflect the complexity of the actual flight scenario, thereby improving the reliability and adaptability of autonomous decision-making.

[0021] Furthermore, the process of generating environmental impact parameters for dynamically adjusting control strategies by calculating obstacle density factors, weather impact factors, and altitude risk factors is as follows: The degree of interference of obstacles on trajectory planning is assessed by using a three-dimensional spatial occupancy grid map, and the obstacle density factor is calculated. The impact of wind speed, visibility, and precipitation intensity information on flight stability is assessed, and weather impact factors are calculated. Assess the risks of altitude selection by combining airspace restrictions and terrain undulations, and calculate the altitude risk factor; Finally, these assessment results were integrated into quantitative environmental impact parameters according to the control strategy of prioritizing obstacle density, secondary weather impact, and supplementing with high risk. Furthermore, the flowchart for calculating the obstacle density factor is shown below. Figure 2 As shown, specifically: a three-dimensional spatial occupancy grid map is established, dividing the flight space into cubic grid cells with a side length of 1m. Each grid cell stores the degree of interference of obstacles on trajectory planning. Point cloud data generated by lidar is mapped to the corresponding grid cells. For each grid cell, the degree of trajectory planning interference is represented as a weighted sum of an altitude factor, a path blocking factor, and a detour cost factor. The altitude factor uses the normalized value of the difference between the obstacle height and the flight altitude to assess whether the obstacle obstructs the trajectory within the flight altitude range. The path blocking factor is calculated... The obstacle's projection occupancy ratio in the grid is used to calculate the ratio of the offset distance required for detours using the detour cost factor. When the grid's interference level exceeds the threshold of 0.7, it is marked as a high-interference state. The basic occupancy rate of the obstacle density factor is calculated by dividing the number of high-interference grids by the total number of effective grids. The spatial distribution variance is analyzed by examining the standard deviation of the coordinate values ​​of high-interference grids on the x, y, and z axes to determine the uneven distribution of high-interference grids in the trajectory planning space. Finally, the obstacle density factor is calculated by multiplying the basic occupancy rate by 0.7 and adding the normalized spatial distribution variance multiplied by 0.3. Furthermore, the weather impact factor is expressed as a weighted sum of the wind speed impact value, visibility impact value, and precipitation impact value, with the final result constrained to the range of 0 to 1 using a hyperbolic tangent function. Specifically, the wind speed impact value is obtained by mapping the current wind speed to the critical wind speed using a sigmoid function; the visibility impact value is the standard visibility value minus the current visibility value, divided by the standard visibility value; and the precipitation impact value is expressed as a weighted sum of the ratio of precipitation intensity to the ratio of snowfall intensity. Furthermore, the calculation process for the altitude risk factor is as follows: Different altitude penalty coefficients are set according to airspace classification. For example, the coefficient is 0.8 for ultra-low altitude (0 to 50 meters), 0.3 for low altitude (50 to 150 meters), 0.1 for medium altitude (150 to 300 meters), and 0.5 for high altitude (above 300 meters). The terrain elevation is calculated by sampling terrain elevation points within a 5 km × 5 km radius around the aircraft, and the ratio of the elevation standard deviation to the average elevation is used to obtain the terrain undulation. When the terrain undulation is less than 0.1, it is considered flat terrain; when it is greater than 0.3, it is considered complex mountainous terrain. Therefore, the altitude risk factor can be expressed as the sum of the current altitude penalty coefficient multiplied by 1 and the terrain undulation. Furthermore, the environmental impact parameters are calculated by multiplying the obstacle density factor by 0.4, the weather impact factor by 0.3, and the high risk factor by 0.3. To ensure that the output is within the range of 0 to 1, the environmental impact parameters are normalized using the hyperbolic tangent function.

[0022] By calculating obstacle density factors, weather impact factors, and altitude risk factors, and integrating them into quantified environmental impact parameters according to priority, it is possible to accurately assess the impact of different factors in the flight environment on trajectory planning and flight stability, providing a scientific basis for dynamically adjusting control strategies, thereby improving the safety and decision-making efficiency of aerial robots in complex environments.

[0023] Based on environmental impact parameters, in a hybrid control framework that includes global path tracking control and local obstacle avoidance control, the control weights of the two are dynamically adjusted through parallel control branches. Under low impact conditions, priority is given to ensuring the optimality of the global trajectory, while under high impact conditions, priority is given to ensuring the safety of local avoidance, thereby achieving adaptive allocation of flight control focus. Furthermore, the process of adaptively allocating flight control focus through parallel control branches in the hybrid control framework is as follows: the small-range perception control branch uses a fine receptive field to extract control information for real-time obstacle avoidance, and the large-range perception control branch uses a wide-area receptive field to extract navigation control information for global path tracking; environmental impact parameters are used to adjust the contribution of each control branch's information to the final control decision. When the environmental impact parameters are low, the weight of the large-range control branch is increased to focus on global navigation control, and when the environmental impact parameters are high, the weight of the small-range control branch is increased to focus on local obstacle avoidance control. Furthermore, the small-range perception control branch employs a fine receptive field to extract control information for real-time obstacle avoidance. It uses a convolutional neural network with 3×3 kernels, 256 input channels, 128 output channels, a receptive field radius of 15m, and a processing frequency of 20Hz. This branch is specifically designed to capture local obstacle features and precise boundary information for local obstacle avoidance control. The large-range perception control branch employs a wide-area receptive field to extract navigation control information for global path tracking. It uses a convolutional neural network with 7×7 kernels, 256 input channels, 128 output channels, a receptive field radius of 100m, and a processing frequency of 5Hz. This branch captures global contextual information such as terrain orientation and route planning for global path tracking control. Furthermore, the dynamic adjustment mechanism for the weights of the parallel control branches is as follows: the environmental impact parameter is used to adjust the contribution of information from the small-scale control branch and the large-scale control branch to the final control decision; the weight of the small-scale control branch is calculated using a sigmoid function, which calculates the product of the difference between the environmental impact parameter and 0.5 and 2; the weight of the large-scale control branch is 1 minus the weight of the small-scale control branch, ensuring that the sum of the two weights is always 1; when the environmental impact parameter is low (less than 0.3), it is a low-impact environment, and the weight of the large-scale control branch is not less than 0.8, focusing on global navigation control; when the environmental impact parameter is high (greater than 0.7), it is a high-impact environment, and the weight of the small-scale control branch is not less than 0.8, focusing on local obstacle avoidance control. Furthermore, the parallel control branch has a dynamic adaptive adjustment mechanism for the sensing field, specifically as follows: the sensing field radius of the small-range sensing control branch and the large-range sensing control branch are related to the base radius, the speed adjustment coefficient, and the complexity adjustment coefficient; the speed adjustment coefficient is a combination of the ratio of the current speed to the maximum speed, a multiplicative coefficient, and an additive coefficient; the complexity adjustment coefficient is a combination of environmental influence parameters, a multiplicative coefficient, and an additive coefficient; the coefficient settings in the speed adjustment coefficient and the complexity adjustment coefficient of the two branches are different, thereby realizing intelligent matching between the sensing range and the flight state, and improving safety and efficiency under different flight conditions.

[0024] By using parallel control branches in the hybrid control framework, local obstacle avoidance and global navigation information can be extracted by small-range and large-range perception control branches respectively. The weight contribution of each branch can be dynamically adjusted according to environmental impact parameters, which enables adaptive allocation of flight control focus. This allows for optimization of global path tracking in low-risk environments and priority to ensure local obstacle avoidance safety in high-risk environments, thereby improving the flexibility and robustness of overall flight control.

[0025] In a hierarchical control architecture consisting of a macro-level mission constraint layer, a meso-level maneuver decision layer, and a micro-level flight control layer, the flight constraints and permitted maneuver behaviors defined at the upper level are transformed into specific aircraft control commands at the lower level through control constraint propagation, and the consistency and safety of the control commands are verified. Furthermore, the hierarchical control architecture, consisting of a macro-level mission constraint layer, a meso-level maneuver decision-making layer, and a micro-level flight control layer, specifically includes: The macro-level mission constraint layer defines the geometric boundaries, no-fly zones, and airspace control rules for flight missions, and uses a geometric-semantic hybrid representation to store airspace control constraints. The meso-level maneuver decision layer provides a set of permissible flight maneuver control actions based on macro-level control constraints, and uses fuzzy production rules to represent maneuver control decisions. The micro-level flight control layer accurately converts selected maneuver actions into direct control parameters for the aircraft. Furthermore, the airspace is divided into 5m×5m grids with a precision of 5m, and the no-fly zone data is stored using an R-tree index structure; airspace control constraints include boundary point coordinate arrays, altitude ranges, constraint types, and constraint boundary values. Furthermore, the flight maneuver control actions provided by the meso-level maneuver decision layer include forward, backward, left turn, right turn, ascent, descent, hovering, precise approach, and rapid withdrawal, with each action associated with 5 to 8 fuzzy rules; Furthermore, the control command update frequency of the micro flight control layer is 100Hz, supporting control parameters such as pitch angle, roll angle, thrust, and yaw rate; By constructing a hierarchical control architecture consisting of a macro-level mission constraint layer, a meso-level maneuver decision layer, and a micro-level flight control layer, which handles mission constraints, maneuver decisions, and control parameter generation respectively, complex flight missions can be decomposed into clear control layers, enabling the orderly transmission from high-level constraints to low-level control. This improves the efficiency of control command generation and the structured and controllable nature of decision-making.

[0026] Furthermore, the process of propagating control constraints step by step into specific aircraft control commands at the lower levels, and verifying the consistency and security of these control commands, is as follows: The macro-level task constraint layer uses the control constraints to filter the maneuver control action options in the meso-level maneuver decision layer, calculates the degree of constraint violation for each control action, and generates control action weights. The micro-level flight control layer maps the maneuver control actions selected by the meso-level maneuver decision layer to a set of initial control parameters based on the current flight control state. Before outputting to the actuators, the control parameters are checked for safety boundaries to ensure that they will not cause the aircraft to stall or exceed the attitude control limits, and safety control corrections are made. Furthermore, the formula for calculating the degree of constraint violation for each control action can be expressed as the absolute value of the difference between the control parameter and the constraint boundary value divided by the constraint deviation tolerance; while the weight of the control action is calculated using an exponential decay function, and the weight of a single control action can be expressed as the result of the exponential decay function calculation of the constraint violation degree of that action divided by the sum of the exponential decay functions of all actions, thereby normalizing it; the ranking result of the control action weights is used as the basis for the meso-level maneuver decision-making layer to select maneuver control actions; Furthermore, the micro-level flight control layer converts the maneuver control actions selected by the meso-level maneuver decision layer into a set of initial control parameters through a parameter mapping table. Taking "forward maneuver" as an example, the mapping parameters include a pitch angle set to -10 degrees, a thrust coefficient set to 0.5, a roll angle maintained at 0 degrees, and a yaw rate set to 0 degrees per second. The parameter mapping takes into account current state corrections, such as reducing the thrust coefficient to 0.3 when the current speed has reached 10 meters per second, and increasing the corresponding roll angle compensation when encountering crosswinds. Before outputting to the actuator, a three-layer safety boundary check is performed: the first layer checks whether the attitude angle is within the safe range, the second layer checks whether the thrust exceeds the rated power, and the third layer checks whether the speed is within the safe range. When any parameter exceeds the safety boundary, the nearest projection method is used to adjust the parameter to within the safety boundary. The safety-corrected parameters are then sent to the flight controller for execution.

[0027] By propagating control constraints step by step into lower-level control commands and verifying their consistency and safety, it is possible to ensure that higher-level constraints effectively constrain lower-level decisions. At the same time, safety boundary checks and corrections can prevent aircraft stalls or attitude loss, thereby improving the reliability of control commands and the safety of the flight process.

[0028] Based on the real-time state vector of the aircraft, the weight coefficients of multiple optimization objectives in the flight control law are dynamically adjusted by the state-weight mapping controller to generate the optimal control command and apply it to the aircraft.

[0029] Furthermore, the process of dynamically adjusting the weight coefficients of multiple optimization objectives in the flight control law through a state-weight mapping controller, based on the real-time state vector of the aircraft, includes: Construct a real-time control state vector that includes the aircraft's own energy state, kinematic state, and mission and environment-related states; where the energy state includes battery percentage, the kinematic state includes flight altitude and current speed, and the mission environment state includes mission urgency score, weather risk level, and obstacle density. The structure of the state-weight mapping controller is as follows: Figure 3As shown, a control state encoder and a control weight generator are used to convert the control state vector into a four-dimensional control weight output that considers safety, efficiency, energy consumption, and accuracy. Specifically, the control state encoder compresses the six-dimensional control state vector through the first, second, and third control processing layers to convert it into a control state feature representation. The control weight generator then converts it into a four-dimensional control weight output through the fourth control processing layer and the output control processing layer. Furthermore, the first control processing layer uses a fully connected layer and the ReLU activation function to extend the control state vector from 6 dimensions to 64 dimensions; the second control processing layer uses a fully connected layer and the ReLU activation function to compress it from 64 dimensions to 32 dimensions; the third control processing layer uses a fully connected layer and the ReLU activation function to compress it from 32 dimensions to 16 dimensions; the fourth control processing layer uses a fully connected layer and the ReLU activation function to compress it from 16 dimensions to 8 dimensions; and the output control processing layer uses the Softmax activation function to reduce the output from 8 dimensions to 4 dimensions, ensuring that the sum of the four control weights—safety, efficiency, energy consumption, and accuracy—is 1. Furthermore, a multi-head attention mechanism is introduced before the first control processing layer of the control state encoder. The process of identifying the importance correlation of different elements in the state vector is as follows: the original 6-dimensional state vector is reorganized into a sequence form and transformed into a state feature matrix through linear transformation; the input state feature matrix is ​​passed through three different linear transformation layers to generate a query matrix, a key matrix, and a value matrix; the correlation weights between state elements are calculated through the multi-head attention layer, and feature fusion is performed through the linear transformation layer to generate an enhanced state feature representation; in multi-objective conflict scenarios, the weight allocation can be more accurately balanced to ensure the rationality of the weight coefficient adjustment of the optimization objective.

[0030] By constructing a real-time state vector that includes energy, kinematics, and mission environment states, and using a state-weighted mapping controller to dynamically adjust the weight coefficients of safety, efficiency, energy consumption, and accuracy, the control law can be optimized based on the real-time state of the aircraft. This allows control commands to better balance multi-objective requirements, thereby improving the adaptability of decision-making and the quality of flight mission completion.

[0031] Furthermore, the dynamic adjustment mechanism of the state-weight mapping controller specifically includes: The state-weighted mapping controller takes the aircraft's real-time control state vector as input and outputs a set of weight coefficients to balance the four control objectives of safety, efficiency, energy consumption, and accuracy. The controller has a built-in safety-priority control mechanism that forcibly increases the weight of the safety control objective when high-risk control states are detected, such as low battery and distance from the landing point or impending collision. At the same time, a smooth switching mechanism for control focus is established, which smooths the control weights through a time sliding window. Furthermore, the minimum weight for each control objective is 0.05 to ensure that all objectives are considered; the maximum weight for a single control objective is 0.70 to avoid excessive bias towards a single objective; situational adaptability constraints include a safety weight of no less than 0.50 when the risk level is greater than 0.8, an efficiency weight of no less than 0.40 when the urgency level is greater than 0.9, and an energy consumption weight of no less than 0.45 when the battery level is less than 0.2; the weight values ​​are not unique. Furthermore, the weight update strategy is as follows: online learning is performed using the Adam optimizer with a learning rate of 0.001, and an experience buffer of the most recent 1000 decisions is maintained; the weight smoothing adopts an exponential moving average, the new weight is the current weight multiplied by 0.8 plus the predicted weight multiplied by 0.2, and the single weight change is limited to 0.1 to avoid drastic changes; Furthermore, the model confidence is comprehensively evaluated by considering the variance of multiple prediction results, the average accuracy of past predictions, and the coverage of the current state in the training data. When the confidence is below 0.7, the state-weight mapping controller switches to conservative mode, increases safety weights, and reduces the exploration rate. Furthermore, the weight coefficients of the control targets also have a context-aware boundary adaptive adjustment mechanism. Specifically, by establishing a context-boundary mapping table, the weight boundary values ​​are dynamically adjusted according to the specific flight context, rather than using a fixed range of weight coefficients. The context-boundary mapping table is as follows: Under emergency rescue missions, the upper limit of efficiency weight is increased to 0.8, and the lower limit of energy consumption weight is reduced to 0.02; under precision measurement missions, the upper limit of accuracy weight is increased to 0.8, and the lower limit of efficiency weight is reduced to 0.03; under long-distance cruise missions, the upper limit of energy consumption weight is increased to 0.8, and the lower limit of accuracy weight is reduced to 0.04; under training flights, each weight maintains the standard boundary. This mechanism enables the weight allocation to better conform to mission characteristics and avoids unreasonable weight constraints.

[0032] By using the dynamic adjustment mechanism of the state-weight mapping controller, combined with the safety priority control and the smooth switching mechanism of control focus, the safety weight can be forcibly increased under high-risk conditions. Smoothing is introduced in the weight adjustment to avoid control instability caused by sudden changes, thereby further enhancing the safety and control smoothness of the aerial robot in complex dynamic environments.

[0033] This embodiment proposes an autonomous decision-making optimization method for aerial robots. It acquires flight environment data, calculates obstacle density factors, weather influence factors, and altitude risk factors, and generates environmental impact parameters for dynamically adjusting the control strategy. A hybrid control framework is constructed, comprising two parallel branches: global path tracking control and local obstacle avoidance control. Control weights are dynamically adjusted based on the environmental impact parameters. A three-layer hierarchical control architecture is established, including a macro-level task constraint layer, a meso-level maneuver decision layer, and a micro-level flight control layer. Upper-level constraints are progressively transformed into lower-level control commands through a control constraint propagation mechanism, and consistency and safety verification are performed. A state-weighted mapping controller is employed to dynamically adjust multi-objective optimization weight coefficients based on real-time state vectors, generating optimal control commands. This invention improves the autonomous decision-making capability and flight safety of aerial robots in dynamic environments.

[0034] Example 2: This embodiment takes a drone inspection task in a certain urban area as an example to illustrate in detail the implementation process of an autonomous decision-making optimization method for aerial robots and demonstrate the application effect of the method in a dynamic environment.

[0035] The system acquires flight environment data of aerial robots and generates environmental impact parameters for dynamically adjusting control strategies by calculating obstacle density factors, weather impact factors, and altitude risk factors. Spatial coordinates and geometric information of obstacles such as buildings, high-voltage lines, and trees within the urban area were collected using airborne lidar. The obstacle density factor was calculated by analyzing the distribution variance of obstacles, resulting in a value of 0.65. The weather impact factor was calculated by weighting the impact values ​​of wind speed, visibility, and rainfall, yielding a value of 0.45. The altitude risk factor, calculated by combining the current altitude penalty coefficient with the terrain undulation, was 0.67.

[0036] By combining the obstacle density factor, weather impact factor, and high risk factor, the environmental impact parameter value is 0.58, which will be used for subsequent control weight adjustments.

[0037] Based on environmental impact parameters, in a hybrid control framework that includes global path tracking control and local obstacle avoidance control, the control weights of the two are dynamically adjusted through parallel control branches. Under low impact conditions, priority is given to ensuring the optimality of the global trajectory, while under high impact conditions, priority is given to ensuring the safety of local avoidance, thereby achieving adaptive allocation of flight control focus. By using parallel control branches, local obstacle information and global navigation information are extracted using small-scale and large-scale control branches respectively. Finally, the branch weights are dynamically adjusted according to the loop influence parameters to achieve flexible flight control information allocation. The weights of the global control branch and the local control branch are 0.62 and 0.85 respectively. In high-risk areas such as dense building areas, the local avoidance weight is increased to 0.9 to ensure the safe detour of the UAV.

[0038] In a hierarchical control architecture consisting of a macro-level mission constraint layer, a meso-level maneuver decision layer, and a micro-level flight control layer, the flight constraints and permitted maneuver behaviors defined at the upper level are transformed into specific aircraft control commands at the lower level through control constraint propagation, and the consistency and safety of the control commands are verified. The macro-level mission constraint layer defines the geometric boundaries, no-fly zones, and airspace control rules for flight missions, and uses a geometric-semantic hybrid representation to store airspace control constraints. The meso-level maneuver decision layer provides a set of permissible flight maneuver control actions based on macro-level control constraints, and uses fuzzy production rules to represent maneuver control decisions. The micro-level flight control layer accurately converts selected maneuver actions into direct control parameters for the aircraft. Based on the real-time state vector of the aircraft, the weight coefficients of multiple optimization objectives in the flight control law are dynamically adjusted by the state-weight mapping controller to generate the optimal control command and apply it to the aircraft.

[0039] The state-weight mapping controller converts the state vector into control weight outputs for safety, efficiency, energy consumption, and accuracy. It has a built-in safety priority mechanism. When the battery level is below 30% or the distance to an obstacle is less than 10m, the safety weight is forcibly increased to 0.65, while other weights are reduced accordingly to ensure that the total weight is 1.

[0040] A time sliding window is used to smooth the weights to avoid control instability caused by sudden changes; finally, optimized weight coefficients are generated, such as safety 0.42, efficiency 0.24, energy consumption 0.19 and accuracy 0.15.

[0041] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An autonomous decision-making optimization method for an aerial robot, characterized in that, include: The system acquires flight environment data of aerial robots and generates environmental impact parameters for dynamically adjusting control strategies by calculating obstacle density factors, weather impact factors, and altitude risk factors. Based on the environmental impact parameters, in a hybrid control framework that includes global path tracking control and local obstacle avoidance control, the control weights of the two are dynamically adjusted through parallel control branches. In low-impact environments, priority is given to ensuring the optimality of the global trajectory, while in high-impact environments, priority is given to ensuring the safety of local avoidance, thereby achieving adaptive allocation of flight control focus. In a hierarchical control architecture consisting of a macro-level mission constraint layer, a meso-level maneuver decision layer, and a micro-level flight control layer, the flight constraints and permitted maneuver behaviors defined at the upper level are transformed into specific aircraft control commands at the lower level through control constraint propagation, and the consistency and safety of the control commands are verified. Based on the real-time state vector of the aircraft, the weight coefficients of multiple optimization objectives in the flight control law are dynamically adjusted by the state-weight mapping controller to generate the optimal control command and apply it to the aircraft.

2. The autonomous decision-making optimization method for an aerial robot according to claim 1, characterized in that, The flight environment data includes obstacle data, meteorological data, and terrain and airspace data.

3. The autonomous decision-making optimization method for an aerial robot according to claim 1, characterized in that, The process of generating environmental impact parameters for dynamically adjusting control strategies by calculating obstacle density factors, weather impact factors, and high risk factors is as follows: The degree of interference of obstacles on trajectory planning is assessed by using a three-dimensional spatial occupancy grid map, and the obstacle density factor is calculated. The impact of wind speed, visibility, and precipitation intensity information on flight stability is assessed, and weather impact factors are calculated. Assess the risks of altitude selection by combining airspace restrictions and terrain undulations, and calculate the altitude risk factor; Finally, these assessment results were integrated into quantitative environmental impact parameters according to the control strategy of prioritizing obstacle density, secondary weather impact, and supplementing with high risk.

4. The autonomous decision-making optimization method for an aerial robot according to claim 1, characterized in that, The process of adaptively allocating flight control focus through parallel control branches in a hybrid control framework is as follows: a small-range perception control branch uses a fine receptive field to extract control information for real-time obstacle avoidance, and a large-range perception control branch uses a wide-area receptive field to extract navigation control information for global path tracking. The environmental impact parameter is used to adjust the contribution of each control branch's information to the final control decision. When the environmental impact parameter is low, the weight of the large-range control branch is increased to focus on global navigation control, and when the environmental impact parameter is high, the weight of the small-range control branch is increased to focus on local obstacle avoidance control.

5. The autonomous decision-making optimization method for an aerial robot according to claim 1, characterized in that, The hierarchical control architecture, consisting of a macro-level mission constraint layer, a meso-level maneuver decision layer, and a micro-level flight control layer, specifically includes: The macro-level mission constraint layer defines the geometric boundaries, no-fly zones, and airspace control rules for flight missions, and uses a geometric-semantic hybrid representation to store airspace control constraints. The meso-level maneuver decision layer provides a set of permissible flight maneuver control actions based on macro-level control constraints, and uses fuzzy production rules to represent maneuver control decisions. The micro-level flight control layer accurately converts selected maneuvers into direct control parameters for the aircraft.

6. The autonomous decision-making optimization method for an aerial robot according to claim 1, characterized in that, The process of propagating control constraints step by step into specific aircraft control commands at the lower levels, and verifying the consistency and security of these control commands, is as follows: The control constraints of the macro-level task constraint layer are used to filter the maneuver control action options in the meso-level maneuver decision layer, calculate the degree of constraint violation for each control action, and generate control action weights. The micro-level flight control layer maps the maneuver control actions selected by the meso-level maneuver decision layer to a set of initial control parameters based on the current flight control state. Before outputting to the actuator, the control parameters are checked for safety boundaries to ensure that they will not cause the aircraft to stall or exceed the attitude control limits, and safety control corrections are made.

7. The autonomous decision-making optimization method for an aerial robot according to claim 1, characterized in that, The process of dynamically adjusting the weight coefficients of multiple optimization objectives in the flight control law based on the real-time state vector of the aircraft, through a state-weight mapping controller, includes: Construct a real-time control state vector that includes the aircraft's own energy state, kinematic state, and mission and environment-related states; where the energy state includes battery percentage, the kinematic state includes flight altitude and current speed, and the mission environment state includes mission urgency score, weather risk level, and obstacle density. The state-weight mapping controller employs a control state encoder and a control weight generator to convert the control state vector into a four-dimensional control weight output that considers safety, efficiency, energy consumption, and accuracy. Specifically, the control state encoder compresses the six-dimensional control state vector sequentially through the first, second, and third control processing layers to convert it into a control state feature representation. The control weight generator then converts it into a four-dimensional control weight output through the fourth control processing layer and the output control processing layer.

8. The autonomous decision-making optimization method for an aerial robot according to claim 7, characterized in that, The dynamic adjustment mechanism of the state-weight mapping controller specifically includes: The state-weight mapping controller takes the real-time control state vector of the aircraft as input and outputs a set of weight coefficients to balance the four control objectives of safety, efficiency, energy consumption and accuracy. The controller has a built-in safety priority control mechanism. When a high-risk control state is detected, such as low battery and being far from the landing point or an impending collision, the weight of the safety control objective is forcibly increased. At the same time, a smooth switching mechanism for control focus is established, which smooths the control weights through a time sliding window.

Images