Fire-fighting unmanned aerial vehicle autonomous navigation method and system for complex building environment

Abstract

The invention discloses a fire-fighting unmanned aerial vehicle autonomous navigation method and system oriented to a complex building environment, and relates to the technical field of unmanned aerial vehicle navigation.The method comprises the steps that firstly, obstacle distribution point cloud data and thermodynamic data are obtained based on a laser radar and other sensors, and a spatial constraint coefficient and a thermal disturbance intensity factor are calculated respectively; therefore, the space trafficability and stability of the flight environment are comprehensively reflected. And then, fusing the global path planning and the local obstacle avoidance to generate a comprehensive environment evaluation parameter, adaptively configuring a weight ratio of the global path planning to the local obstacle avoidance according to the comprehensive environment evaluation parameter, and carrying out weighted fusion on instructions output by different algorithms so as to generate a smooth and safe flight control instruction. According to the invention, the dynamic adjustment of the navigation strategy can be realized according to the change of the building environment, the obstacle avoidance safety in the complex environment is ensured, the continuity and efficiency of the overall path planning are also considered, and the autonomous flight stability and task execution reliability of the fire-fighting unmanned aerial vehicle in the high-risk building environment are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) navigation technology, specifically to an autonomous navigation method and system for firefighting UAVs in complex building environments. Background Technology

[0002] With the development of drone technology and intelligent control algorithms, firefighting drones are increasingly widely used at fire scenes for tasks such as fire reconnaissance, temperature monitoring, and assisting in firefighting. However, in complex building environments such as high-rise buildings and underground passages, drone flight navigation faces multiple challenges. These environments are characterized by narrow interior spaces, dense obstacles, and intense thermal airflow disturbances and smoke obstruction, making it difficult for traditional path planning algorithms based on lidar or visual perception to simultaneously meet the requirements of global path optimization and real-time obstacle avoidance.

[0003] Current navigation technologies often employ fixed threshold switching or simple fusion methods to schedule between global planning and local obstacle avoidance algorithms, lacking flexible quantitative analysis of the surrounding environment. This leads to policy lag or flight instability when environmental conditions change. Especially in fire scenes, changes in heat flux gradients can cause UAV attitude jitter and sensing errors, further weakening the stability of path control. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an autonomous navigation method and system for firefighting drones in complex building environments.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows:

[0006] In a first aspect, the present invention discloses an autonomous navigation method for firefighting drones in complex building environments, comprising the following steps:

[0007] Acquire point cloud data of obstacle distribution and thermodynamic data in the flight environment;

[0008] Based on the point cloud data, the spatial constraint coefficient is calculated by analyzing the passable space and obstacle distribution density represented by the point cloud data.

[0009] Based on the aforementioned thermodynamic data, the thermal disturbance intensity factor is calculated by analyzing the gradient change of the ambient temperature value represented by the thermodynamic data and combining it with the spectral characteristics of the attitude jitter data.

[0010] The spatial constraint coefficient and the thermal disturbance intensity factor are fused to generate comprehensive environmental assessment parameters;

[0011] The comprehensive environmental assessment parameters are compared with at least one preset threshold, and corresponding global planning weights and local obstacle avoidance weights are configured for the current environment based on the comparison relationship, wherein the sum of the global planning weights and the local obstacle avoidance weights is a fixed value;

[0012] Based on the configured global planning weights and local obstacle avoidance weights, the global path command generated by the global planning algorithm according to the mission objective and the local obstacle avoidance command generated by the local obstacle avoidance algorithm according to point cloud data and thermodynamic data are weighted and fused to generate the final flight control command.

[0013] Execute the final flight control command to control the drone's flight.

[0014] Secondly, this invention discloses an autonomous navigation system for firefighting drones designed for complex building environments, comprising:

[0015] The data acquisition module is used to acquire point cloud data of obstacle distribution and thermodynamic data in the flight environment;

[0016] The spatial constraint calculation module is used to calculate the spatial constraint coefficient based on the point cloud data by analyzing the passable space and obstacle distribution density represented by the point cloud data.

[0017] The disturbance intensity calculation module is used to calculate the thermal disturbance intensity factor based on the thermodynamic data by analyzing the gradient change of the ambient temperature value represented by the thermodynamic data and combining the spectral characteristics of the attitude jitter data.

[0018] The evaluation parameter fusion module is used to fuse the spatial constraint coefficient and the thermal disturbance intensity factor to generate comprehensive environmental evaluation parameters;

[0019] The weight configuration module is used to compare the comprehensive environmental assessment parameters with at least one preset threshold, and configure corresponding global planning weights and local obstacle avoidance weights for the current environment based on the comparison relationship, wherein the sum of the global planning weights and the local obstacle avoidance weights is a fixed value;

[0020] The flight command generation module is used to weight and fuse the global path command generated by the global planning algorithm based on the mission objective and the local obstacle avoidance command generated by the local obstacle avoidance algorithm based on point cloud data and thermodynamic data, based on the configured global planning weight and the local obstacle avoidance weight, to generate the final flight control command.

[0021] The flight control module is used to execute the final flight control commands and control the flight of the UAV.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0023] 1. By calculating the spatial constraint coefficient and thermal disturbance intensity factor, and generating comprehensive environmental assessment parameters accordingly, the priority of global path planning and local obstacle avoidance strategies can be weighed in real time under complex building structures and dynamic thermal airflow environments, so that UAVs can maintain flight safety while taking into account mission completion efficiency.

[0024] 2. Compared with existing navigation methods that rely on fixed parameters or single sensor data, this solution demonstrates stronger robustness and environmental adaptability in typical fire scenarios such as smoke obstruction, hot air updrafts, and narrow passages. It can effectively reduce navigation deviation and instability risks, and significantly improve the autonomous navigation performance and mission execution reliability of firefighting drones in complex building environments. Attached Figure Description

[0025] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein:

[0026] Figure 1 This is a flowchart of the method of the present invention;

[0027] Figure 2 This is a data flow diagram of the present invention;

[0028] Figure 3 This is a system architecture diagram of the present invention. Detailed Implementation

[0029] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0030] Application Overview

[0031] Currently, firefighting drones are increasingly used in building fire rescue scenarios. However, the dense distribution of obstacles and intense thermal airflow disturbances in complex building structures make it easy for traditional navigation methods that rely on lidar or visual SLAM to suffer from problems such as path planning distortion, obstacle avoidance delay, and unstable flight attitude, making it difficult to ensure flight safety and mission continuity.

[0032] This invention aims to address the problem of unmanned aerial vehicles (UAVs) being unable to achieve stable and adaptive navigation in complex building and thermal disturbance environments, thereby improving the environmental adaptability and autonomous decision-making capabilities of UAVs in fire scenes. To this end, this invention proposes a dynamic navigation weight control method that integrates spatial constraint characteristics and thermal disturbance features. This method first acquires obstacle distribution point cloud data and thermodynamic data of the flight environment. Based on the point cloud data, it analyzes the accessibility characteristics of the building space to calculate the spatial constraint coefficient, and based on the thermodynamic data combined with attitude jitter spectrum characteristics, it calculates the thermal disturbance intensity factor. Then, it fuses the two data to generate a comprehensive environmental assessment parameter, and dynamically configures the global planning weight and local obstacle avoidance weight according to this parameter, achieving an adaptive balance in navigation decision-making while keeping the sum of the two constant. Finally, it weights and fuses the global path command and the local obstacle avoidance command to generate flight control commands, achieving stable flight control of the UAV within complex buildings.

[0033] Through the aforementioned technical means, this invention achieves joint modeling and real-time weight allocation of spatial complexity and thermal disturbance, enabling firefighting drones to dynamically adjust their navigation strategies according to the actual environment, maintaining high robustness and stability even in smoke, heat waves, and obstacle-heavy scenarios. This method can significantly improve the autonomous navigation accuracy and mission execution reliability of firefighting drones in complex building fire scenes, possessing significant practical value and promotional significance.

[0034] like Figure 1 , Figure 2 As shown, this application proposes an autonomous navigation method for firefighting drones in complex building environments, comprising the following steps:

[0035] The system acquires point cloud data and thermodynamic data of obstacle distribution in the flight environment. Point cloud data can be obtained using a lidar sensor, while thermodynamic data can be acquired using an infrared thermal imager or temperature sensor array. The point cloud data can be processed using 3D reconstruction algorithms to extract obstacle information from the environment. Thermodynamic data may include information such as ambient temperature distribution and temperature gradient changes. Furthermore, attitude jitter data can be acquired using an inertial measurement unit (IMU) to analyze the stability of the UAV under thermal airflow disturbances.

[0036] Based on the point cloud data, a spatial constraint coefficient is calculated by analyzing the passable space and obstacle distribution density represented by the point cloud data. When calculating the spatial constraint coefficient, the volume of the passable space within a preset fan-shaped area in front of the UAV can be calculated using geometric estimation methods, and then weighted in conjunction with the obstacle point cloud density. Specifically, the volume of the passable space can be achieved through voxelization or 3D mesh generation, while the obstacle distribution density can be determined by statistically analyzing the number of point clouds per unit volume. A dynamic weight adjustment mechanism can be introduced into the calculation of the spatial constraint coefficient to adapt to navigation requirements under different environmental conditions.

[0037] Based on the aforementioned thermodynamic data, the thermal disturbance intensity factor is calculated by analyzing the gradient change of the ambient temperature value represented by the thermodynamic data and combining it with the spectral characteristics of the attitude jitter data. The calculation of the thermal disturbance intensity factor can be based on the temperature gradient change and the spectral characteristics of the attitude jitter. The temperature gradient can be obtained through linear regression analysis within a sliding time window, and corrected by incorporating the standard deviation of the temperature difference between adjacent sampling points. The spectral characteristics of the attitude jitter can be obtained through Fast Fourier Transform (FFT) analysis and compared with a reference stationary spectrum to calculate the spectral deviation characteristic. A sensitivity coefficient can be introduced into the calculation of the thermal disturbance intensity factor to adjust the degree of response to temperature changes and attitude jitter.

[0038] The spatial constraint coefficient and the thermal disturbance intensity factor are fused to generate comprehensive environmental assessment parameters. These parameters can be generated through weighted fusion or enhanced calculation. A dynamic adjustment mechanism can be introduced into the calculation of the comprehensive environmental assessment parameters to adapt to assessment needs under different environmental conditions. When the spatial constraint coefficient or thermal disturbance intensity factor exceeds a danger threshold, enhanced calculation is used; otherwise, weighted fusion is used.

[0039] The comprehensive environmental assessment parameters are compared with at least one preset threshold, and based on the comparison, corresponding global planning weights and local obstacle avoidance weights are configured for the current environment. The sum of the global planning weights and the local obstacle avoidance weights is a fixed value. The configuration of the global planning weights and local obstacle avoidance weights can be based on the comparison with the preset thresholds. Specifically, multiple threshold intervals can be set, and the weight allocation method is determined according to the interval in which the comprehensive environmental assessment parameters fall. An interpolation function can be introduced into the weight calculation to achieve a smooth transition and avoid flight instability caused by sudden weight changes.

[0040] Based on the configured global planning weights and local obstacle avoidance weights, the global path command generated by the global planning algorithm according to the mission objective and the local obstacle avoidance command generated by the local obstacle avoidance algorithm based on real-time perception data are weighted and fused to generate the final flight control command. The generation of the final flight control command can be achieved through vector weighted fusion. The velocity and heading angle information in the global path command and the local obstacle avoidance command can be weighted and fused separately to obtain the final flight control command. The fused command can be dynamically limited to ensure flight safety.

[0041] Execute the final flight control command to control the drone's flight.

[0042] This method achieves flexible scheduling of global path planning and real-time obstacle avoidance by quantitatively analyzing environmental conditions. In complex building environments, it effectively addresses challenges such as confined spaces, dense obstacles, and thermal airflow disturbances. Compared to existing technologies that use fixed threshold switching or simple fusion methods, this method dynamically adjusts navigation strategies based on environmental conditions, improving the stability and adaptability of flight control. By comprehensively evaluating spatial constraints and thermal disturbance factors, this method can effectively address local environmental changes while ensuring optimal global path planning, thus enhancing the autonomous navigation capabilities of firefighting drones in complex building environments.

[0043] This application also proposes a specific implementation method for configuring global planning weights and local obstacle avoidance weights for the current environment after comparing comprehensive environmental assessment parameters with preset thresholds. Specifically, this includes:

[0044] Determine the relationship between the comprehensive environmental assessment parameters and the preset first and second thresholds, where the second threshold is greater than the first threshold;

[0045] If the comprehensive environmental assessment parameters are lower than or equal to the first threshold, then a high fixed value is configured for the global planning weight, and a low fixed value is configured for the local obstacle avoidance weight.

[0046] If the comprehensive environmental assessment parameter is higher than or equal to the second threshold, then a high fixed value is configured for the local obstacle avoidance weight, and a low fixed value is configured for the global planning weight.

[0047] If the comprehensive environmental assessment parameters are between the first threshold and the second threshold, then the parameters are normalized based on their relative positions within that range, and the specific values ​​of the global planning weight and the local obstacle avoidance weight are continuously calculated using a preset interpolation function.

[0048] The first threshold and the second threshold are denoted as follows: , These two thresholds will affect the entire environmental assessment parameters (denoted as...). The numerical range of ) is divided into three decision intervals with clear physical meaning:

[0049] Low-risk area ( This range represents a low level of environmental constraint and threat to drone navigation. It typically corresponds to open outdoor spaces without strong thermal disturbances or large indoor spaces. Within this range, the navigation strategy prioritizes efficiency. Therefore, a high fixed value (e.g., 0.8 or 0.9) is configured for the global planning weight, while a low fixed value (e.g., 0.2 or 0.1) is configured for the local obstacle avoidance weight. This allows the drone to consistently fly along the globally optimal path, avoiding only sudden obstacles not marked on the global map.

[0050] High-risk area ( This range represents an extremely complex and dangerous environment. It typically corresponds to a narrow corridor filled with obstacles, an area of ​​intense thermal airflow, or a combination of both. Within this range, the navigation strategy prioritizes survival. Therefore, the strategy is reversed, assigning a high fixed value (e.g., 0.8 or 0.9) to the local obstacle avoidance weights and a low fixed value (e.g., 0.2 or 0.1) to the global planning weights. In this case, the UAV's primary task is to utilize real-time sensor data to ensure it avoids collisions with any obstacles or hazards; the global path serves only as a general reference direction.

[0051] Transition interval ( This interval represents a continuous change from safe to dangerous, from open to narrow. Using fixed weight values ​​within this interval is unreasonable, as it would cause the UAV's navigation strategy to abruptly change when crossing the threshold, leading to flight instability. To address this issue, this invention employs a continuous calculation method based on normalization and interpolation functions to achieve a smooth transition of weight values.

[0052] For this stage, the specific calculation process for the global planning weight and the local obstacle avoidance weight is as follows:

[0053] First, calculate the normalization parameters. : Normalized parameters It reflects the current environment's relative position within the transition zone.

[0054] Then, a preset interpolation function is used. Applied to normalized parameters To continuously calculate the current global planning weights. and local obstacle avoidance weights ;

[0055] ; indicates From a high fixed value in the low-risk range Low fixed value transitioning to high-risk range ;

[0056] ; indicates global planning weight and local obstacle avoidance weights The sum is a fixed value.

[0057] For interpolation functions The following options are available:

[0058] The formula is a cubic polynomial, called the "smooth step" function; it is based on the normalized parameters. The first derivative is zero when the value is 0 or 1, which means that the start and end points of the weight change curve are smoothly connected by tangents in the transition interval, completely eliminating any tiny jumps during policy switching, thus achieving the smoothest and most stable drone behavior transition.

[0059] Interpolation functions can also be used. Select linear interpolation function: Weight follows Linear change.

[0060] Therefore, this technical solution achieves dynamic and smooth adjustment of global planning and local obstacle avoidance weights by introducing threshold comparison and continuous interpolation mechanisms. Compared with existing technologies that use fixed threshold switching or simple fusion, this solution can respond more accurately to changes in environmental conditions, avoiding policy lag or flight instability. In complex built environments, especially with thermal airflow disturbances, continuously adjusted weight configurations help maintain flight control stability while balancing the needs of global path optimization and real-time obstacle avoidance.

[0061] By employing normalization and continuous interpolation functions, this method achieves dynamic and smooth adjustment of global planning weights and local obstacle avoidance weights in response to changes in environmental conditions. Compared to existing technologies that use fixed threshold switching, which can lead to abrupt policy changes, this approach generates gradual weight allocation results based on continuous changes in environmental assessment parameters, thereby improving flight control stability. Particularly in complex scenarios such as fire scenes with thermal disturbances and varying obstacle density, this calculation method avoids flight command oscillations caused by weight jumps, ensuring stable flight of the UAV in transitional environmental states.

[0062] Furthermore, this application also proposes a calculation process for the spatial constraint coefficients. Specifically, it includes the following steps:

[0063] Extract the point cloud data and calculate the volume of passable space within a preset fan-shaped area in front of the drone using geometric estimation. The obstacle distribution density is calculated by statistically analyzing the number of obstacle point clouds per unit volume space within the fan-shaped region. The angle of the preset sector area can be set from 90 degrees to 180 degrees, and the radius can be dynamically adjusted according to the size and flight speed of the drone.

[0064] The spatial constraint coefficient is calculated using a weighted calculation formula. :

[0065] ,in, This is a preset reference value for the maximum passable space volume, which can be set according to different building types. to ; This is a preset reference value for the maximum obstacle point cloud density, which can be set to 1000. Up to 5000 ; , These are the preset weighting coefficients, and they satisfy... Its specific value can be determined through optimization using experimental data, for example... , .

[0066] As a preferred implementation method, the passable space volume The calculation can be performed using a voxelization method, dividing the sector into several voxel units, counting the number of voxels not occupied by obstacles, and multiplying the number by the volume of a single voxel. Obstacle distribution density. The density can be calculated using the kernel density estimation method, which smooths the point cloud data using a Gaussian kernel function to improve the accuracy of density calculation. and The settings can take into account parameters such as the drone's minimum turning radius and obstacle avoidance reaction time to ensure that the calculated spatial constraint coefficient can accurately reflect the complexity of the flight environment.

[0067] This technical solution, by quantitatively analyzing the spatial characteristics of the fan-shaped area in front of the UAV and comprehensively considering two key factors—accessible space and obstacle density—can accurately assess the complexity of the flight environment. Compared with existing technologies, the weighted calculation method allows for flexible adjustment of the weights of different factors, adapting to the characteristics of different architectural environments; the preset fan-shaped area setting focuses on the critical area where the UAV is about to fly, improving computational efficiency; and the introduction of a maximum reference value achieves normalization under different environmental conditions, making the spatial constraint coefficients comparable. Therefore, this solution provides a reliable environmental assessment basis for subsequent path planning and obstacle avoidance decisions.

[0068] This application also proposes a calculation process for the thermal disturbance intensity factor, which, through the fusion analysis of multi-source thermodynamic data, accurately quantifies the dynamic impact of hot airflow on the flight stability of UAVs in a fire environment; specifically including:

[0069] First, key thermodynamic data are extracted from airborne sensors, which are divided into ambient temperature value sequences and attitude jitter data spectrum.

[0070] Ambient temperature value sequence: continuously acquiring a predetermined number of bits at a fixed sampling frequency (e.g., most recent). The ambient temperature readings (per sampling period) are used to form a time series data set. This series is used to analyze the dynamic trend of temperature changes.

[0071] Attitude jitter data spectrum: The raw data of UAV attitude angles (such as roll angle and pitch angle) or angular velocities measured by the inertial measurement unit are acquired synchronously and transformed from the time domain to the frequency domain through fast Fourier transform to obtain the spectrum of the current attitude jitter data. This spectrum reflects the energy distribution of the disturbances experienced by the UAV at different frequencies.

[0072] Calculate the temperature change gradient based on the ambient temperature value sequence. The deviation characteristic of the current attitude jitter data from the reference stationary spectrum is calculated based on the spectrum of the current attitude jitter data. ;

[0073] Among them, temperature change gradient It can effectively capture the overall trend of temperature changes and avoid noise interference caused by instantaneous fluctuations; deviation characteristic quantity The larger the value, the stronger the abnormal disturbance experienced by the drone.

[0074] To comprehensively assess the impact of thermal disturbance, the following weighted formula is used to fuse the two characteristic quantities into a unified thermal disturbance intensity factor. :

[0075] ;

[0076] in, The preset maximum temperature gradient reference value is a threshold set based on fire scenario experience, used to monitor temperature change gradients. Normalization is performed to transform it into a dimensionless form with a range of... Nearby relative values. When Approaching or exceeding At this point, the temperature gradient is generally considered to have reached a dangerous level.

[0077] The preset maximum spectral deviation reference value is also an empirical threshold used to measure the deviation characteristic. Normalize the values ​​to convert them into comparable dimensionless values.

[0078] and The preset weighting coefficients represent the contributions of temperature gradient and attitude spectrum deviation to the final evaluation. They satisfy the following relationship: By adjusting these two coefficients, you can adapt to different aircraft models or mission requirements. For example, for drones with extremely high attitude stability requirements, you can set a larger value. This value makes the system more sensitive to attitude jitter.

[0079] Through the above calculation process, this invention achieves comprehensive perception and quantification of thermal disturbances from their "cause" (temperature gradient) to their "result" (attitude jitter). Normalizing and weighting these two different dimensions of information results in a final thermal disturbance intensity factor. This provides a reliable and comprehensive indicator that reflects the overall threat posed by the thermal environment to flight stability, offering precise data for the intelligent adjustment of subsequent UAV navigation strategies.

[0080] Furthermore, this application also proposes a temperature change gradient. The calculation process is as follows:

[0081] Set a sliding time window to continuously acquire the environmental temperature value sequence of the most recent predetermined number of sampling points within the window; the width of the sliding time window can be dynamically adjusted according to the actual application scenario. For example, a narrower window can be used in high temperature and strong convection environment to improve response speed, while a wider window can be used in relatively stable temperature areas to improve measurement stability.

[0082] Linear regression analysis was performed on the environmental temperature value series to obtain the slope of temperature change over time. As a temperature gradient The baseline value; linear regression analysis can be performed using the least squares method, and the standard deviation of the temperature difference score can be calculated and updated in real time using the sliding standard deviation algorithm.

[0083] Calculate the temperature difference between adjacent sampling points in the ambient temperature value sequence, and then determine the standard deviation of the temperature difference. ;

[0084] The final temperature gradient is calculated using the formula. :

[0085] ;

[0086] in, This is a preset sensitivity coefficient, typically set to a range of 0.1-0.5. The standard deviation is a preset benchmark value, which can be obtained from historical data statistics.

[0087] Therefore, this technical solution achieves precise quantification of temperature gradient changes in a thermally disturbed environment by combining the trend characteristics and local fluctuation features of temperature changes. Compared with methods that simply use linear regression slope or difference standard deviation, this composite calculation method can simultaneously reflect the overall trend of temperature changes and the degree of local abrupt changes, thus more accurately characterizing the complex thermal flow disturbance state in a fire environment. Specifically, when there are high-frequency fluctuations in the temperature sequence, the standard deviation... The introduction of this method effectively enhances the sensitivity of gradient detection; while when the temperature exhibits a stable trend, the linear regression slope k ensures the accuracy of the basic measurement. This dual-consideration mechanism allows the calculation results to adapt to sudden changes in heat flow at fire scenes while maintaining measurement stability under normal conditions.

[0088] Furthermore, this application also proposes that the process for generating comprehensive environmental assessment parameters is as follows:

[0089] Determine the spatial constraint coefficient Is it greater than the first danger threshold? or thermal disturbance intensity factor Is it greater than the second danger threshold? Spatial constraint coefficient Reflecting the spatial complexity of the environment, it is calculated using point cloud data and characterizes the combined impact of passable space and obstacle density; thermal disturbance intensity factor. The degree of thermodynamic disturbance in the environment is reflected and calculated using temperature gradient changes and attitude jitter spectrum characteristics. First danger threshold. Second danger threshold Separate settings are provided for spatial constraints and thermal disturbances to identify high-risk environmental conditions.

[0090] If the judgment result is yes, then the comprehensive environmental assessment parameters are calculated using the following formula. :

[0091] ,in The preset enhancement coefficient;

[0092] If the judgment result is negative, the comprehensive environmental assessment parameters are calculated using the following formula. :

[0093] ,in , The weighting coefficients are preset, and .

[0094] Specifically, the enhancement coefficient The contribution of secondary risk factors is adjusted, and its value is recommended to be between 0.1 and 0.3. The weighting coefficient is used to balance the effects of spatial constraints and thermal disturbances, and a fixed proportion is used in a safe environment.

[0095] When the environment is in a high-risk state, adopting a nonlinear fusion strategy can effectively amplify the impact of the dominant risk factors and prevent the underestimation of a single risk factor. For example, at a fire scene, if thermal disturbances suddenly intensify, even if the spatial constraint coefficient is low, this approach can promptly improve the comprehensive risk assessment level by selecting the maximum value.

[0096] Compared to existing technologies that use fixed weights or simple switching, this solution, through a dual judgment mechanism and differentiated fusion strategy, can more accurately reflect the dynamic risk combinations in complex building environments, providing a more precise input basis for subsequent path planning weight allocation.

[0097] Therefore, this technical solution addresses the problem of inaccurate risk assessment in complex and ever-changing environments using traditional methods by establishing an environmental state classification mechanism and a differentiated parameter fusion strategy. Its core lies in automatically switching the fusion algorithm based on the degree of environmental hazard, ensuring balanced assessment under normal conditions while achieving risk focus under emergency conditions. This adaptive assessment mechanism significantly improves the UAV's response speed and control stability to changes in environmental conditions.

[0098] Furthermore, this application proposes that the final flight control command generation process intelligently and smoothly fuses global path commands based on long-term mission objectives and local obstacle avoidance commands based on real-time perception, according to weights determined by the current environmental assessment, to output stable and reliable control commands. This process specifically includes the following steps:

[0099] Extract the desired velocity vector from the global path instructions output by the global planning algorithm. and expected heading angle This command guides the UAV to move efficiently toward the final mission objective, with its velocity vector pointing toward the next waypoint on the global path.

[0100] Extract the obstacle avoidance velocity vector from the local obstacle avoidance command output by the local obstacle avoidance algorithm. and obstacle avoidance heading angle The instruction is designed to enable the drone to immediately avoid currently perceived obstacles, with its velocity vector pointing towards a collision-free safe space.

[0101] Calculate the fusion velocity vector using the velocity fusion formula. :

[0102] ;

[0103] in, For global planning weights, Local obstacle avoidance weights;

[0104] The velocity vector itself has directionality. When the drone is in an open environment ( high, (low), fusion result Mainly refers to The drone flies along a global path in the direction indicated by the obstacle. When approaching an obstacle... (increase) The influence of the fused velocity vector is enhanced, and its direction is usually directed away from the obstacle. By naturally deviating from the original global path, the drone can avoid obstacles by circumventing them. This method ensures a continuous and smooth change in the drone's direction of movement.

[0105] Calculate the blended heading angle using the heading angle blending formula. :

[0106] ;

[0107] The heading angle is a scalar angle. Weighted averaging effectively combines two potentially conflicting angle commands (e.g., a global command to head east while a local obstacle avoidance command to head north) into a compromise heading command that matches the current weight allocation. This ensures smooth changes in the UAV's nose direction, avoiding flight instability or jitter caused by drastic and frequent switching between two command sources. The denominator in the formula is the sum of the weights, ensuring that the calculated result is a valid angle weighting value under any weight configuration.

[0108] Fusion velocity vector Integrated heading angle These are combined into the final flight control commands.

[0109] Therefore, this technical solution effectively addresses the challenge of simultaneously achieving global path stability and real-time obstacle avoidance in complex building environments through a hierarchical command extraction and dynamic fusion mechanism. This weight-based fusion mechanism enables the UAV to behave intelligently and naturally in complex environments, much like an experienced pilot, significantly improving the reliability and adaptability of autonomous navigation in high-risk scenarios such as firefighting.

[0110] Furthermore, this application proposes a dynamic flight limiting process, a crucial step in ensuring the flight safety of UAVs in complex environments. Its core lies in dynamically calculating and limiting the maximum flight speed of the UAV based on real-time environmental risk assessment, thereby significantly improving survivability in hazardous environments while maintaining basic maneuverability. Specifically, this includes:

[0111] Obtain current comprehensive environmental assessment parameters and spatial constraint coefficient ;

[0112] The maximum permissible speed value is calculated using the dynamic limiting function. :

[0113] ;

[0114] The dynamic limiting function dynamically determines the maximum permissible speed by combining the product of an exponential decay term and a linear suppression term, taking into account both environmental risks and spatial constraints. .

[0115] in, The preset maximum speed value is the maximum speed that the drone is allowed to reach under ideal and safe conditions, and serves as the benchmark for speed limits.

[0116] As a whole, it is considered an environmental risk assessment item. This is the environmental assessment attenuation coefficient; this is an exponential attenuation model. When considering comprehensive environmental assessment parameters... When the temperature increases (e.g., when thermal disturbances are intense or the fire is fierce). The value decreases rapidly. This means that the more dangerous the environment, the faster the maximum permissible speed decays. The exponential model ensures a sensitive response at the onset of risk and suppresses the speed to a very low level under high risk, greatly improving safety. (Coefficient) The severity of the decay was controlled.

[0117] As a spatial constraint suppression term This is the spatial constraint suppression coefficient. This represents the maximum reference value for the spatial constraint coefficient; this is a linear inhibition model. It directly reflects the physical limitation of speed on the degree of spatial congestion in the environment. The narrower the space and the denser the obstacles (i.e., the higher the spatial constraint coefficient), the greater the speed limitation. The larger the value of this term (the larger it is), the smaller its value, thus affecting... It produces a stronger inhibition. This determines the strength of the spatial constraints on velocity. When and When this value is zero, it means that in extremely confined spaces, the speed will be limited to near zero, forcing the drone to hover or move at extremely slow speeds.

[0118] Calculate the fusion velocity vector Current module length This indicates the actual speed at which the current command requires the drone to fly.

[0119] Determine the current module length Does it exceed the current maximum allowed speed value? ;

[0120] If the judgment result is yes, then the velocity vector is limited according to the following formula:

[0121] ; indicates that the direction of the velocity vector remains unchanged, only its magnitude is scaled to . . It is the new velocity vector after amplitude limiting. This ensures that the drone's intention to avoid obstacles or track paths is preserved while it decelerates.

[0122] If the judgment result is negative, the original velocity vector will remain unchanged. This indicates that the current commanded velocity is within a safe range, and the system will maintain the fused velocity vector. No change, output directly.

[0123] This technical solution employs dynamic limiting to adjust the maximum flight speed of the UAV in real time according to environmental conditions, thereby achieving safe and stable flight control in complex environments. Compared with existing technologies, this solution avoids the problems of low flight efficiency or insufficient safety caused by fixed speed limits, and can flexibly adapt to flight requirements under different environmental conditions. Specifically, in areas with confined spaces or strong thermal disturbances, dynamic limiting can effectively reduce flight speed, minimizing collision risks and attitude instability; in more relaxed environments, flight speed can be appropriately increased to improve mission execution efficiency. Thus, this solution ensures flight safety while balancing flight efficiency and stability.

[0124] Furthermore, such as Figure 3 The diagram shown is the system architecture diagram of this application. This application also proposes an autonomous navigation system for firefighting drones in complex building environments. Through the collaborative work of seven core functional modules, a complete, closed-loop autonomous navigation and control loop is constructed. This system achieves fully autonomous decision-making from environmental perception to flight control, specifically addressing the challenges of the dual complexity of spatial structure and thermodynamic environment in firefighting scenarios. It includes a data acquisition module, a spatial constraint calculation module, a disturbance intensity calculation module, an evaluation parameter fusion module, a weight configuration module, a flight command generation module, and a flight control module.

[0125] The data acquisition module is the system's perception front end, responsible for collecting real-time environmental information about the UAV's location. Specifically, it acquires point cloud data of obstacle distribution in the surrounding environment through onboard LiDAR or depth cameras to describe the environment's geometry; simultaneously, it collects thermodynamic data, including ambient temperature readings and raw data reflecting the UAV's attitude jitter, through a temperature sensor array and inertial measurement unit. This module provides raw data input for all subsequent analyses.

[0126] The spatial constraint calculation module is responsible for processing geometric spatial information and quantifying the congestion level of the environment. It receives point cloud data from the data acquisition module and estimates the volume of passable space within a pre-defined fan-shaped area in front of the UAV using specific algorithms (such as 3D voxel mesh analysis or convex hull calculation), while simultaneously calculating the obstacle point cloud density within that area. Finally, based on these two features, a single, dimensionless spatial constraint coefficient is output using a weighted calculation formula. A larger coefficient value indicates a narrower space and a higher flight difficulty.

[0127] The disturbance intensity calculation module is responsible for processing thermodynamic information and quantifying the threat level of thermal airflow to flight stability. It receives thermodynamic data from the data acquisition module. On one hand, it performs linear regression analysis or difference calculations on continuous temperature sequences to obtain the temperature change gradient; on the other hand, it performs a fast Fourier transform on attitude jitter data to obtain the spectrum and compares it with a reference spectrum to calculate the spectral deviation characteristic. Finally, these two characteristic quantities are normalized and weighted to output a thermal disturbance intensity factor. The larger the factor value, the stronger the thermal disturbance.

[0128] The evaluation parameter fusion module is responsible for generating a unified environmental assessment conclusion. It receives spatial constraint coefficients from the spatial constraint calculation module and thermal disturbance intensity factors from the disturbance intensity calculation module. A fusion strategy combines these two factors, representing physical space risk and thermodynamic risk respectively, into a comprehensive environmental assessment parameter. This parameter is the system's overall score for the current environmental "hazard," serving as the direct basis for subsequent navigation strategy switching.

[0129] The weight configuration module receives comprehensive environmental assessment parameters from the evaluation parameter fusion module and compares them with preset thresholds. Based on this comparison, the module dynamically configures global planning weights and local obstacle avoidance weights for the current environment according to preset rules. The sum of these two weights is a fixed value, and their relative magnitudes directly determine the UAV's tendency to "adhere to the global plan" and "respond to local obstacle avoidance".

[0130] The flight command generation module receives weight values ​​from the weight configuration module, as well as global path commands generated by the global planning algorithm based on the mission objective, and local obstacle avoidance commands generated by the local obstacle avoidance algorithm based on real-time point cloud and thermodynamic data. The core function of this module is to weight and fuse these two types of commands to generate a final flight control command that balances efficiency and safety.

[0131] The flight control module is the system's execution terminal. It receives the final flight control commands (usually including the target velocity vector and heading angle) from the flight command generation module and converts them into control signals for the underlying servos and motors, driving the UAV to complete corresponding flight maneuvers, such as forward movement, turning, and ascent / descend.

[0132] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. An autonomous navigation method for firefighting drones in complex building environments, characterized by: Includes the following steps: Acquire point cloud data and thermodynamic data of obstacle distribution in the flight environment; Based on the point cloud data, the spatial constraint coefficient is calculated by analyzing the passable space and obstacle distribution density represented by the point cloud data. Based on the aforementioned thermodynamic data, the thermal disturbance intensity factor is calculated by analyzing the gradient change of the ambient temperature value represented by the thermodynamic data and combining it with the spectral characteristics of the attitude jitter data. The spatial constraint coefficient and the thermal disturbance intensity factor are fused to generate comprehensive environmental assessment parameters; The comprehensive environmental assessment parameters are compared with at least one preset threshold, and corresponding global planning weights and local obstacle avoidance weights are configured for the current environment based on the comparison relationship, wherein the sum of the global planning weights and the local obstacle avoidance weights is a fixed value; Based on the configured global planning weights and local obstacle avoidance weights, the global path command generated by the global planning algorithm according to the mission objective and the local obstacle avoidance command generated by the local obstacle avoidance algorithm according to point cloud data and thermodynamic data are weighted and fused to generate the final flight control command. Execute the final flight control command to control the drone's flight.

2. The autonomous navigation method for firefighting drones in complex building environments according to claim 1, characterized in that: Based on comparison relationships, appropriate global planning weights and local obstacle avoidance weights are assigned to the current environment, including: Determine the relationship between the comprehensive environmental assessment parameters and the preset first and second thresholds, where the second threshold is greater than the first threshold; If the comprehensive environmental assessment parameters are lower than or equal to the first threshold, then a high fixed value is configured for the global planning weight, and a low fixed value is configured for the local obstacle avoidance weight. If the comprehensive environmental assessment parameter is higher than or equal to the second threshold, then a high fixed value is configured for the local obstacle avoidance weight, and a low fixed value is configured for the global planning weight. If the comprehensive environmental assessment parameters are between the first threshold and the second threshold, then the parameters are normalized based on their relative positions within that range, and the specific values ​​of the global planning weight and the local obstacle avoidance weight are continuously calculated using a preset interpolation function.

3. The autonomous navigation method for firefighting drones in complex building environments according to claim 2, characterized in that: The specific calculation process for the global planning weight and the local obstacle avoidance weight is as follows: Calculate normalization parameters : ,in The current values ​​of the comprehensive environmental assessment parameters are as follows. The value of the first threshold. This is the value of the second threshold. The global planning weights are calculated using an interpolation function. : ,in, The high-order fixed value configured for the global planning weights, The low-order fixed value configured for the global planning weights, To normalize parameters The interpolation function for the variable is given by the following expression: ； Local obstacle avoidance weights The calculation formula is: 。 4. The autonomous navigation method for firefighting drones in complex building environments according to claim 1, characterized in that: The calculation process for the spatial constraint coefficient is as follows: Extract the point cloud data and calculate the volume of passable space within a preset fan-shaped area in front of the drone using geometric estimation. The obstacle distribution density is calculated by statistically analyzing the number of obstacle point clouds per unit volume space within the fan-shaped region. ; The spatial constraint coefficient is calculated using a weighted calculation formula. : ,in, This is a preset reference value for the maximum passable space volume. This is a preset reference value for the maximum obstacle point cloud density. , These are the preset weighting coefficients, and they satisfy... .

5. The autonomous navigation method for firefighting drones in complex building environments according to claim 1, characterized in that: The calculation process for the thermal disturbance intensity factor includes: Based on the thermodynamic data, extract the ambient temperature value sequence of a predetermined sampling period and the spectrum of the current attitude jitter data; Calculate the temperature change gradient based on the ambient temperature value sequence. The deviation characteristic of the current attitude jitter data from the reference stationary spectrum is calculated based on the spectrum of the current attitude jitter data. ; The thermal disturbance intensity factor is calculated using a weighted calculation formula. : ,in, This is the preset maximum temperature gradient reference value. The preset maximum spectral deviation reference value, and The preset weighting coefficients, and satisfy the following conditions: .

6. The autonomous navigation method for firefighting drones in complex building environments according to claim 5, characterized in that: The temperature change gradient The calculation process is as follows: Set a sliding time window and continuously acquire the environmental temperature value sequence of the most recent predetermined number of sampling points within the window; Linear regression analysis was performed on the environmental temperature value series to obtain the slope of temperature change over time. As a temperature change gradient The baseline value; Calculate the temperature difference between adjacent sampling points in the ambient temperature value sequence, and then determine the standard deviation of the temperature difference. ; The final temperature gradient is calculated using the formula. : ； in, The preset sensitivity coefficient, This is the preset standard deviation benchmark value.

7. The autonomous navigation method for firefighting drones in complex building environments according to claim 1, characterized in that: The process for generating the comprehensive environmental assessment parameters is as follows: Determine the spatial constraint coefficient Is it greater than the first danger threshold? or thermal disturbance intensity factor Is it greater than the second danger threshold? ; If the judgment result is yes, then the comprehensive environmental assessment parameters are calculated using the following formula. : ,in The preset enhancement coefficient; If the judgment result is negative, the comprehensive environmental assessment parameters are calculated using the following formula. : ,in , The weighting coefficients are preset, and .

8. The autonomous navigation method for firefighting drones in complex building environments according to claim 1, characterized in that: The process of generating the final flight control command is as follows: Extract the desired velocity vector from the global path command. and expected heading angle ; Extract the obstacle avoidance velocity vector from the local obstacle avoidance command. and obstacle avoidance heading angle ; Calculate the fusion velocity vector using the velocity fusion formula. : ； Calculate the blended heading angle using the heading angle blending formula. : ； in, For global planning weights, Local obstacle avoidance weights; Fusion velocity vector Integrated heading angle These are combined into the final flight control commands.

9. The autonomous navigation method for firefighting drones in complex building environments according to claim 8, characterized in that: The fusion velocity vector Dynamic limiting is performed after calculation, specifically including: Obtain current comprehensive environmental assessment parameters and spatial constraint coefficient ; The maximum permissible speed value is calculated using the dynamic limiting function. : ;in, The preset base maximum speed value, For environmental assessment attenuation coefficient, This is the spatial constraint suppression coefficient. This is the maximum reference value for the spatial constraint coefficient; Calculate the fusion velocity vector Current module length ; Determine the current module length Does it exceed the current maximum allowed speed value? ; If the judgment result is yes, then the velocity vector is limited according to the following formula: , The velocity vector after amplitude limiting; If the judgment result is negative, then the original velocity vector remains unchanged.

10. An autonomous navigation system for firefighting drones in complex building environments, characterized in that: include: The data acquisition module is used to acquire point cloud data of obstacle distribution and thermodynamic data in the flight environment; The spatial constraint calculation module is used to calculate the spatial constraint coefficient based on the point cloud data by analyzing the passable space and obstacle distribution density represented by the point cloud data. The disturbance intensity calculation module is used to calculate the thermal disturbance intensity factor based on the thermodynamic data by analyzing the gradient change of the ambient temperature value represented by the thermodynamic data and combining the spectral characteristics of the attitude jitter data. The evaluation parameter fusion module is used to fuse the spatial constraint coefficient and the thermal disturbance intensity factor to generate comprehensive environmental evaluation parameters; The weight configuration module is used to compare the comprehensive environmental assessment parameters with at least one preset threshold, and configure corresponding global planning weights and local obstacle avoidance weights for the current environment based on the comparison relationship, wherein the sum of the global planning weights and the local obstacle avoidance weights is a fixed value; The flight command generation module is used to weight and fuse the global path command generated by the global planning algorithm based on the mission objective and the local obstacle avoidance command generated by the local obstacle avoidance algorithm based on point cloud data and thermodynamic data, based on the configured global planning weight and the local obstacle avoidance weight, to generate the final flight control command. The flight control module is used to execute the final flight control commands and control the flight of the UAV.

Images