Prism-based joint positioning method and system for UAV and robot dog in tunnels

Abstract

The present invention discloses a method and system for jointly positioning a drone and a robot dog in a tunnel based on prism reference points, belonging to the field of information processing technology. The method comprises: constructing a tunnel prism reference point array, obtaining an initial reflection signal and preprocessing it to obtain a processed signal set; determining a signal dynamic threshold and, based on the signal dynamic threshold, obtaining a reference point set; obtaining reflection signal data from the drone and the robot dog using the reference point set and extracting reflection signal characteristics; determining the initial position distribution of the drone and the robot dog based on the reflection signal characteristics; obtaining dynamic reflection data from the drone and the robot dog based on the initial position distribution, processing the dynamic reflection data to obtain a smoothed position sequence; calculating a relative motion trend based on the smoothed position sequence and the motion difference characteristics of the drone and the robot dog; determining the position consistency of the drone and the robot dog based on the relative motion trend; and determining the real-time positioning coordinates of the drone and the robot dog based on the position consistency.

Description

Technical Field

[0001] The present invention belongs to the field of information processing technology, and in particular relates to a method and system for jointly positioning a UAV and a robot dog in a tunnel based on a prism reference point. Background Art

[0002] In complex underground environments such as tunnels, the joint positioning technology of drones and robot dogs is of vital importance to ensuring efficient collaborative operations. Research in this field has not only promoted the application of intelligent robotic systems in unmanned environments, but also provided key technical support for scenarios such as disaster relief and underground exploration. However, existing positioning methods face significant limitations in tunnel environments. Traditional GPS-based positioning technology almost completely fails when the signal is blocked in the tunnel, while solutions based on inertial navigation or visual SLAM are prone to accumulated errors during long-term operation or in complex dynamic environments, making it difficult to meet the requirements of high precision and stability. In addition, single-device positioning makes it difficult to achieve multi-device collaboration, limiting mission efficiency and coverage.

[0003] In the joint positioning method based on prism reference points, the core challenge lies in how to efficiently and stably track the prism reflection signal and convert it into accurate positioning data. First, the deployment of the prism and signal capture need to remain stable in the complex environment of the tunnel to avoid signal loss due to light interference or device movement. Secondly, the motion characteristics of the drone and the robot dog are significantly different. How to achieve collaborative positioning of the two through the prism reference point to ensure real-time and consistency is a technical difficulty. Finally, the system needs to quickly process the relative position changes of the reflected signal under limited computing resources to meet real-time positioning requirements. These unresolved technical factors have led to fluctuations in positioning accuracy and insufficient efficiency in multi-device collaboration, which in turn affected the execution of complex tasks.

[0004] Therefore, how to achieve efficient signal tracking and collaborative positioning of drones and robot dogs through prism reference points in a tunnel environment has become a key issue that needs to be solved urgently. Summary of the Invention

[0005] The present invention aims to address the deficiencies of the prior art and proposes a method for jointly positioning a UAV and a robot dog in a tunnel based on a prism reference point, comprising the following steps:

[0006] S1. Constructing a tunnel prism reference point array, obtaining an initial reflection signal based on the prism reference point array, preprocessing the initial reflection signal to obtain a processed signal set; determining a signal dynamic threshold based on the tunnel light interference distribution characteristics, and obtaining a reference point set based on the processed signal set and the signal dynamic threshold;

[0007] S2. Acquire reflection signal data of the UAV and the robot dog through the reference point set and extract reflection signal features; determine the initial position distribution of the UAV and the robot dog based on the reflection signal features;

[0008] S3. Acquire dynamic reflection data of the UAV and the robot dog based on the initial position distribution, and process the dynamic reflection data to obtain a smooth position sequence;

[0009] S4. Calculating a relative motion trend based on the smoothed position sequence and the motion difference characteristics of the drone and the robot dog; and determining the position consistency of the drone and the robot dog based on the relative motion trend;

[0010] S5. Determine the real-time positioning coordinates of the drone and the robot dog based on the position consistency.

[0011] Further preferably, S2 includes the following steps:

[0012] S21. Acquire reflection signal data of the drone and the robot dog, perform feature extraction on the reflection signal data to obtain reflection signal features; and determine distribution characteristics of signal strength and signal distance based on the reflection signal features;

[0013] S22. Calculate preliminary position estimates of the drone and the robot dog using triangulation based on the distribution characteristics of the signal strength and the signal distance;

[0014] S23 , correcting the preliminary position estimate, calculating a matching degree between the corrected preliminary position estimate and the tunnel environment, and obtaining the initial position distribution.

[0015] Further preferably, S3 includes the following steps:

[0016] S31, obtaining the dynamic reflection data of the drone and the robot dog, and using a Kalman filter algorithm to fuse the time series to determine a position estimate;

[0017] S32. Determine dynamic change characteristics caused by movement based on the position estimate; and determine the continuity of the movement trajectory based on the dynamic change characteristics.

[0018] S33. When the continuity of the moving trajectory meets the preset threshold, the position estimation value is optimized to obtain an updated position sequence; when the continuity of the moving trajectory does not meet the preset threshold, the process returns to S31;

[0019] S34. Calculate trajectory smoothness based on the updated position sequence, and adjust parameters of a Kalman filter algorithm based on the trajectory smoothness, thereby obtaining the smoothed position sequence.

[0020] Further preferably, the method for calculating the relative motion trend based on the smoothed position sequence and the motion difference characteristics of the drone and the robot dog comprises the following steps:

[0021] S41, extracting motion trajectory points of the drone and the robot dog according to the smoothed position sequence, and determining motion characteristics of the drone and the robot dog respectively based on the motion trajectory points;

[0022] S42, calculating the displacement difference between the drone and the robot dog at each time point based on the motion characteristics, thereby obtaining a velocity vector;

[0023] S43. Calculate a direction vector based on the angular change of the velocity vector to obtain the movement directions of the drone and the robot dog; and obtain the relative movement trend based on the movement directions.

[0024] Further preferably, the method for determining the position consistency of the drone and the robot dog based on the relative motion trend includes:

[0025] S44, obtaining a preliminary position result set based on the acquired real-time reflection signal data and the relative motion trend;

[0026] S45. Extracting reference coordinate data from the preliminary position result set to determine the relative position distribution of the drone and the robot dog;

[0027] S46. Based on the relative position distribution, perform signal fusion processing on the real-time reflection signal to obtain an adjusted position solution result; determine the deviation between the adjusted position solution result and the relative motion trend; if the deviation exceeds a set deviation threshold, re-perform signal fusion processing; if the deviation is below the set deviation threshold, obtain updated position data of the drone and the robot dog, and thereby obtain the real-time position relationship between the drone and the robot dog;

[0028] S47. Based on the real-time position relationship, the motion trend is optimized to obtain an optimized position solution result; and based on the optimized position solution result, the position consistency of the drone and the robot dog is determined.

[0029] Further preferably, the method for determining the position consistency between the drone and the robot dog includes:

[0030] S471, extracting signal tracking deviation values ​​from the optimized position solution result to obtain a deviation value sequence;

[0031] S472: Compare the values ​​in the deviation value sequence with a preset deviation threshold. If the deviation threshold is exceeded, determine the signal segment that needs to be resampled and obtain resampled data; otherwise, determine that the positions of the drone and the robot dog are consistent.

[0032] S473, performing feature extraction on the resampled data to obtain a signal feature set; and performing smoothing processing on the signal features to obtain smoothed feature data;

[0033] S474 , calculating the position correction amount based on the smoothing feature data to obtain position correction data; inputting the position correction data into S471 to extract the signal tracking deviation value.

[0034] The present invention also proposes a joint positioning system of a UAV and a robot dog in a tunnel based on a prism reference point, which is used to implement the above method, comprising: a reference point acquisition module, a feature extraction module, a processing module, a position calculation module, and a positioning module;

[0035] The reference point acquisition module is used to construct a tunnel prism reference point array, obtain an initial reflection signal based on the prism reference point array, pre-process the initial reflection signal to obtain a processed signal set; determine a signal dynamic threshold based on the tunnel light interference distribution characteristics, and obtain a reference point set based on the processed signal set and the signal dynamic threshold;

[0036] The feature extraction module obtains the reflected signal data of the UAV and the robot dog through the reference point set and extracts the reflected signal features; based on the reflected signal features, determines the initial position distribution of the UAV and the robot dog;

[0037] The processing module is used to obtain dynamic reflection data of the UAV and the robot dog based on the initial position distribution, and process the dynamic reflection data to obtain a smooth position sequence;

[0038] The position calculation module is used to calculate the relative motion trend based on the smooth position sequence and the motion difference characteristics of the drone and the robot dog; and determine the position consistency of the drone and the robot dog based on the relative motion trend;

[0039] The positioning module is used to determine the real-time positioning coordinates of the drone and the robot dog based on the position consistency.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention obtains initial reflection signal data through a preset prism reference point array and determines the coordinates of stable reference points based on the light interference characteristics. The signal data is processed using triangulation to preliminarily determine the device location distribution. The Kalman filter algorithm is used to fuse time series data to obtain smooth position updates. The speed direction vector is calculated based on the differences in device motion characteristics to determine the relative motion trend. The consistent position is determined by fusing real-time signal changes. When the deviation exceeds the threshold, particle filter resampling is used to optimize positioning. Finally, the processing frequency is adjusted according to the computing resource limitations, and the real-time positioning coordinates are output. The present invention solves the problem of collaborative positioning of multiple devices in the complex environment of tunnels and improves positioning accuracy and real-time performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0043] Figure 1 This is a flow chart of a method for jointly positioning a UAV and a robot dog in a tunnel based on prism reference points proposed in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] Example 1:

[0047] like Figure 1 As shown, this embodiment proposes a method for jointly positioning a UAV and a robot dog in a tunnel based on a prism reference point, including the following steps:

[0048] S1. Construct a tunnel prism reference point array, obtain an initial reflection signal based on the prism reference point array, preprocess the initial reflection signal to obtain a processed signal set; combine the tunnel light interference distribution characteristics to determine the signal dynamic threshold, and obtain a reference point set based on the processed signal set and the signal dynamic threshold.

[0049] Specifically, a prism reference point array is preset, and the high reflective characteristics of the prism are used to capture light signals in the environment. In one implementation, in a tunnel scenario, a prism array is deployed on the tunnel wall, and a prism is set every 5 meters to collect the reflected light emitted by the laser rangefinder. The initial reflection signal obtained contains effective reflection signals and background light noise. Therefore, a time domain analysis method is used to process the initial reflection signal through a time window, separate the effective signal and noise, and obtain a processed signal set. If there is high-frequency interference in the processed signal set, a filter is used for filtering. For example, in a tunnel, electrical equipment may introduce interference above 50KHz. The low-frequency part can be retained by setting a cutoff frequency for the low-pass filter. The filtered signal is smoother and the noise spikes are suppressed.

[0050] Taking into account the distribution of light interference in tunnel environments, the dynamic threshold is determined by calculating the mean and variance of the signal strength. For example, if the mean of the filtered signal strength is 100 units and the variance is 15 units, the dynamic threshold can be set to the mean plus twice the variance, or 130 units. This method adapts to light variations and avoids misjudgments using a fixed threshold. The use of a dynamic threshold improves the robustness of signal screening.

[0051] Within the dynamic threshold range, reflection points with signal strength exceeding the dynamic threshold are screened one by one to obtain a benchmark point set. During this screening, timestamps can be incorporated to exclude unusually dense interference points, further improving the reliability of the benchmark points. Cluster analysis is used to group adjacent reflection points within the benchmark point set to obtain a stable benchmark point set. For example, using the DBSCAN clustering algorithm, with a neighborhood radius of 0.5 meters and a minimum number of points of 3, 200 benchmark points are divided into 10 clusters, each representing a stable reflection area. Cluster analysis effectively eliminates isolated points and enhances the stability of the benchmark point set.

[0052] Geometric positioning algorithms can be used to calculate the spatial coordinates of reference points. For example, triangulation, using the relative positions of a laser rangefinder and a prism, can calculate the coordinates of a reference point to be (10, 2, 1.5) meters. The final coordinate set contains the three-dimensional coordinates of all reference points, forming a high-precision positioning framework within the tunnel. This positioning accuracy can reach the centimeter level, providing reliable data support for tunnel monitoring.

[0053] Through a progressive process of signal acquisition, filtering, threshold screening, clustering, and positioning, data quality and positioning accuracy are ensured. In tunnel environments with complex lighting conditions, the combination of dynamic thresholding and clustering analysis significantly improves anti-interference capabilities. The resulting coordinate set can be used for tunnel deformation monitoring or automated navigation, significantly enhancing safety and efficiency.

[0054] S2. Obtain the reflected signal data of the drone and the robot dog through the reference point set and extract the reflected signal features; based on the reflected signal features, determine the initial position distribution of the drone and the robot dog.

[0055] A further implementation is that S2 includes the following steps:

[0056] S21. Obtain reflection signal data of the drone and the robot dog, perform feature extraction on the reflection signal data, and obtain reflection signal features; determine the distribution characteristics of signal strength and signal distance based on the reflection signal features.

[0057] The reflected signal data from the drone and robot dog is obtained using a set of reference point coordinates. This reflected signal data typically contains information such as time and intensity. Feature extraction technology is used to process the reflected signal data and determine the distribution characteristics of signal strength and signal distance.

[0058] S22. Based on the distribution characteristics of signal strength and signal distance, the triangulation method is used to calculate the preliminary position estimates of the drone and the robot dog.

[0059] S23. Correct the preliminary position estimate, calculate the matching degree between the corrected preliminary position estimate and the tunnel environment, and obtain an initial position distribution.

[0060] For example, if a reference point is 50 meters from the drone, the round-trip time of the reflected signal is approximately 0.33 microseconds, and the signal strength is weakened by reflection from the tunnel walls. This signal characteristic provides basic data for subsequent analysis. Feature extraction techniques are used to process the reflected signals and determine the distribution characteristics of signal strength and signal distance. If the reflected signals come from multiple directions, feature extraction identifies the primary angle range, such as 30 to 45 degrees. The distance is then calculated based on the signal propagation time. Assuming the speed of light, a time difference of 0.5 microseconds corresponds to approximately 75 meters. This distribution characteristic helps determine the signal source direction and range, thereby improving positioning accuracy. If the drone signal angles measured at reference points A and B are 40 and 50 degrees, respectively, and the distance between the two points is 100 meters, geometric relationships can be used to preliminarily estimate that the drone is approximately 60 meters ahead of the line connecting the two points. Similarly, the robot dog's position can be inferred from another set of signal data. This method is simple and intuitive and applicable to the regular environment of a tunnel. The initial estimate is adjusted based on the constraints of the tunnel environment to obtain corrected position distribution data. Constraints include tunnel width and height. For example, if the tunnel is only 10 meters wide, but the initial estimate shows the drone at 15 meters, clearly out of range, the position can be corrected to within the boundary, for example, to 9 meters. This correction better matches the actual environment, avoids significant errors, and helps improve positioning reliability. If the deviation between the corrected position distribution data and the reference point coordinates exceeds a preset threshold (for example, if the threshold is 2 meters and the corrected position deviates by 3 meters), the reflected signal characteristics are recalculated. Reanalyzing the reflected signal intensity distribution reveals that interference at the tunnel corners has caused data offset. After adjustment, the deviation can be reduced to within 1.5 meters. This iterative optimization ensures data consistency. The reflected signal characteristics are compared with the tunnel environment. For example, if a section of the tunnel has a metal wall, the reflected signal intensity is abnormally high, consistent with the corresponding environmental results. This is then determined to be the final position. Key point coordinates are extracted from the final position distribution to determine the relative position of the two devices in the tunnel environment. For example, if the drone's coordinates are (50, 5, 3) and the robot dog's are (45, 6, 1), it can be inferred that the drone is 5 meters ahead of the robot dog, with a height difference of 2 meters. This relative relationship can be used to coordinate the actions of two devices, such as a drone guiding a robot dog to avoid obstacles. This analysis provides a clear spatial basis for collaborative tasks in tunnels.

[0061] S3. Based on the initial position distribution, the dynamic reflection data of the UAV and the robot dog are obtained, and the dynamic reflection data are processed to obtain a smooth position sequence.

[0062] In a further implementation, S3 includes the following steps:

[0063] S31. Obtain dynamic reflection data of the drone and the robot dog, and use the Kalman filter algorithm to fuse the time series to determine the position estimate.

[0064] A drone or robot dog transmits ultrasonic signals in a tunnel. A receiving device records the duration of the signals from emission to return, forming a time series. For example, if each measurement interval is 0.1 seconds and continuous acquisition is performed for 10 seconds, 100 data points are obtained. These data points reflect the distance relationship between the drone or robot dog and its surroundings, forming a rough location distribution.

[0065] For example, if the initial predicted position is (11, 21, 5.5) and the measured value is (12, 22, 6), the fusion results in a smoothed position estimate of (11.8, 21.8, 5.9). This effectively reduces noise interference and improves position stability. The position estimate can be used to analyze dynamic change characteristics.

[0066] S32. Determine dynamic change characteristics caused by movement based on the position estimation value; and judge the continuity of the movement trajectory based on the dynamic change characteristics.

[0067] For example, if 10 consecutive position estimates show a drone moving in a straight line with an average displacement of approximately 0.5 meters per second, its movement is considered continuous. Continuity can be assessed by calculating the distance difference between adjacent points. If each displacement difference is less than 1 meter, the condition is considered met. This continuity assessment can confirm the authenticity of the device's movement and avoid misjudgments.

[0068] S33. When the continuity of the moving trajectory meets the preset threshold, the position estimation value is optimized to obtain an updated position sequence; when the continuity of the moving trajectory does not meet the preset threshold, the process returns to S31.

[0069] When trajectory continuity meets the requirements, the dynamic reflection data is low-pass filtered to remove high-frequency noise, retain the main trends, and generate an updated position sequence. This can more clearly demonstrate the dynamic changes in position distribution, such as the trajectory of a drone moving steadily along the axis of a tunnel.

[0070] S34. Calculate trajectory smoothness based on the updated position sequence, and adjust parameters of the Kalman filter algorithm based on the trajectory smoothness to obtain a smooth position sequence.

[0071] Extract trends from the updated position sequence and calculate smoothness. Smoothness can be measured using the change in curvature. If the curvature remains below 0.1, the trajectory is smooth. Based on the smoothing results, adjust the Kalman filter parameters, such as increasing the process noise covariance, to generate a smoothed position sequence.

[0072] S4. Calculate the relative motion trend based on the smooth position sequence and the motion difference characteristics of the UAV and the robot dog; and determine the position consistency of the UAV and the robot dog based on the relative motion trend.

[0073] In a further embodiment, the method for calculating the relative motion trend based on the smoothed position sequence and the motion difference characteristics of the drone and the robot dog includes the following steps:

[0074] S41 , extracting motion trajectory points of the drone and the robot dog according to the smooth position sequence, and determining motion characteristics of the drone and the robot dog respectively based on the motion trajectory points.

[0075] S42. Calculate the displacement difference between the drone and the robot dog at each time point based on the motion characteristics, and then obtain the velocity vector.

[0076] S43. Calculate the direction vector based on the angular change of the velocity vector to obtain the movement directions of the drone and the robot dog; and obtain the relative movement trend based on the movement directions.

[0077] The trajectory points of the drone and robot dog are extracted from the smoothed position sequence, and their coordinates at time t1 are recorded. The displacement difference between the drone and robot dog from time t1 to t2 is then calculated. For example, at t1, the coordinates of the drone and robot dog are (10, 20, 5) and (8, 15, 0), respectively. At t2, the drone and robot dog have moved to (12, 22, 3) and (10, 17, 0), respectively, with displacement differences of (2, 2, -2) and (2, 2, 0), respectively. This displacement difference indicates that the drone dives downward, while the robot dog maintains horizontal movement.

[0078] The velocity component is calculated from the displacement difference. The magnitude of the velocity component reflects the speed of the drone and robot dog's movement, while the angle indicates direction. If the drone's velocity vector is (2, 2, -2) and the robot dog's is (1, 1, 0), calculating the direction vector based on the angle change reveals that the drone dives downward while the robot dog maintains horizontal movement. This difference helps determine the relative motion relationship. If the difference in the modulus of the velocity and direction vectors exceeds a preset threshold, the characteristic difference is calculated using the inner product of the vectors. Assuming the inner product result shows that the drone and robot dog's motion directions are at a 45-degree angle, they are not completely synchronized. This degree of deviation quantifies the kinematic independence of the two devices and provides a basis for subsequent analysis.

[0079] Linear regression is used to fit relative trends. If the data from five time points shows a gradually decreasing displacement difference from 5 to 2, the two devices are considered to be approaching each other. The stability of this trend helps predict collaboration or conflict between devices. When determining the future motion relationship based on the rate of change of the relative trend, if the rate of change approaches 0, it indicates that the two devices may enter a stable following state. For example, the robot dog adjusts its pace to follow the drone's descent point, or the drone hovers and waits for the robot dog to approach. This prediction can optimize path planning and improve collaborative efficiency. If the rate of change is positive and increasing, such as from 0.1 to 0.5, it indicates that the two devices are moving apart, and the control strategy may need to be adjusted to maintain the mission objective. This analysis method ensures a comprehensive understanding of the dynamic relationship by comprehensively considering trajectory, velocity, and direction, effectively supporting real-time monitoring and adjustment of inter-device motion. The coordinated analysis of direction and velocity vectors also provides a reference for navigation in complex environments and improves the robustness of motion trajectories.

[0080] In a further embodiment, the method for determining the position consistency of the drone and the robot dog based on the relative motion trend includes:

[0081] S44. Obtain a preliminary position result set based on the acquired real-time reflection signal data and relative motion trend.

[0082] For example, when a drone hovers at a certain altitude, it detects changes in the intensity of the signal reflected from the ground. Meanwhile, when a robot dog moves on the ground, its signal is affected by the undulating terrain. After incorporating the relative motion trends, the initial position solution is derived from the signal time difference and the direction of movement. For example, if the drone is hovering at an altitude of 10 meters and the robot dog is moving eastward at 2 meters per second on the ground, the initial solution might indicate a horizontal distance of 15 meters.

[0083] S45. Extract reference coordinate data from the preliminary position result set to determine the relative position distribution of the UAV and the robot dog.

[0084] When extracting reference coordinate data from the initial position solution, the drone's position can be set to the coordinate origin (0,0,10) and the robot dog's position to (15,0,0). This establishment of reference coordinates helps to intuitively reflect the distribution of the two in three-dimensional space. Signal fusion methods can adjust the coordinates based on the relative position distribution, combining the strength and angle of the reflected signal. For example, a decrease in signal strength indicates an increase in distance, while an angular deviation indicates a change in direction, thereby correcting for initial solution deviations.

[0085] S46. Based on the relative position distribution, the real-time reflection signal is subjected to signal fusion processing to obtain the adjusted position solution result; the deviation between the adjusted position solution result and the relative motion trend is determined. When the deviation exceeds the set deviation threshold, the signal fusion processing is performed again; when the deviation is lower than the set deviation threshold, the updated position data of the UAV and the robot dog are obtained, and then the real-time position relationship of the UAV and the robot dog is obtained.

[0086] During actual operation, there may be multi-sensor data, such as the gyroscope of a robot dog and the GPS of a drone. Signal fusion processing can effectively reduce the error of a single signal source and ensure that the solution results are closer to reality.

[0087] S47. Based on the real-time position relationship, the Kalman filter algorithm is used to optimize the motion trend to obtain the optimized position solution result. Based on the optimized position solution result, the position consistency of the drone and the robot dog is determined.

[0088] In a further embodiment, the method for determining the position consistency between the drone and the robot dog includes:

[0089] S471. Extract the signal tracking deviation value from the optimized position solution result, and use statistical analysis method to obtain the deviation value sequence.

[0090] S472. Compare the values ​​in the deviation value sequence with a preset deviation threshold. When the deviation threshold is exceeded, determine the signal segment that needs to be resampled and obtain resampled data; otherwise, determine that the positions of the drone and the robot dog are consistent.

[0091] S473. Perform feature extraction on the resampled data to obtain a signal feature set; and perform smoothing on the signal features to obtain smoothed feature data.

[0092] S474 , calculating the position correction amount based on the smoothing feature data to obtain position correction data; inputting the position correction data into S471 to extract the signal tracking deviation value.

[0093] In a collaborative positioning scenario involving a drone and a robot dog, assuming the signal tracking deviation is derived from the timestamp differences in the reflected signals, the deviation might be 5 milliseconds at one moment and 8 milliseconds at another. Statistical analysis is used to calculate the deviation sequence. Specifically, a sliding window averaging method can be used to generate a sequence of 10 consecutive deviation values, such as 5, 6, 7, 8, 6, 5, 4, 3, 2, and 1 millisecond. If the preset threshold is 7 milliseconds, then 8 in the sequence exceeds the threshold, indicating that the signal may be subject to interference and requires further processing. This method intuitively reflects signal stability and helps quickly locate problematic segments. For deviation values ​​exceeding the threshold, the resampled signal segment is determined. The specific interval can be determined by comparing timestamps. For example, a deviation value of 8 corresponds to the signal segment from the 4th to the 6th second.

[0094] This judgment relies not only on the magnitude of the deviation but also on the continuity of the signal to avoid misjudging a single outlier. Resampling improves signal quality, effectively reducing errors in subsequent analysis. Resampling the reflected signal using a particle filter algorithm divides the signal into multiple particle groups, each representing a possible signal state.

[0095] For example, for the signal segment from the 4th to 6th second, 100 particles are generated, each simulating the signal shape under different noise levels, ultimately selecting the optimal solution. This method preserves the diverse characteristics of the signal and ensures that the resampled data is closer to reality. When extracting feature points from the resampled signal data, one can focus on the peaks and valleys of the signal. For example, a delay feature might manifest as a signal delay of 2 milliseconds, while an amplitude feature might manifest as an intensity change of 10 decibels. The formation of a set of signal features provides a reliable foundation for subsequent smoothing processing. The Kalman filter algorithm is used to smooth the feature data. This smoothing process reduces the impact of noise and improves data consistency. Position corrections are calculated using the smoothed feature data, and can be extrapolated based on the changing trends of delay and amplitude. For example, a 0.2 millisecond reduction in delay might correspond to a 0.5 meter eastward shift in the drone's position. This correction directly optimizes the positioning result.

[0096] S5. Determine the real-time positioning coordinates of the drone and the robot dog based on position consistency.

[0097] Considering the limited capabilities of the processing equipment, the signal processing frequency is optimized and adjusted, and the position correction data is calculated to obtain the real-time positioning coordinates.

[0098] Through the above-mentioned joint positioning method, the accurate position information of the UAV and the robot dog can be obtained. During tunnel operations, the robot dog can serve as a landing pad for the UAV, while giving full play to their respective advantages. The UAV can obtain high-precision image data or laser data of the tunnel face or other engineering work surfaces with high altitude requirements. The combination of robot dogs and UAVs has huge application value in the field of engineering inspection.

[0099] Example 2:

[0100] This embodiment provides a joint positioning system for a UAV and a robot dog in a tunnel based on a prism reference point, which is used to implement the method provided in the above embodiment, including: a reference point acquisition module, a feature extraction module, a processing module, a position calculation module and a positioning module.

[0101] The reference point acquisition module is used to construct a tunnel prism reference point array, obtain the initial reflection signal based on the prism reference point array, preprocess the initial reflection signal to obtain a processed signal set; combine the tunnel light interference distribution characteristics to determine the signal dynamic threshold, and obtain the reference point set based on the processed signal set and the signal dynamic threshold.

[0102] The feature extraction module obtains the reflection signal data of the UAV and the robot dog through the reference point set and extracts the reflection signal features; based on the reflection signal features, the initial position distribution of the UAV and the robot dog is determined.

[0103] The processing module is used to obtain the dynamic reflection data of the UAV and the robot dog based on the initial position distribution, and process the dynamic reflection data to obtain a smooth position sequence.

[0104] The position calculation module is used to calculate the relative motion trend based on the smooth position sequence and the motion difference characteristics of the drone and the robot dog; and to determine the position consistency of the drone and the robot dog based on the relative motion trend.

[0105] The positioning module is used to determine the real-time positioning coordinates of the drone and the robot dog based on position consistency.

[0106] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A joint positioning method for UAVs and robot dogs in tunnels based on prism reference points, characterized in that: The following steps are involved: S1. Constructing a tunnel prism reference point array, obtaining an initial reflection signal based on the prism reference point array, and preprocessing the initial reflection signal to obtain a processed signal set; Combined with the distribution characteristics of tunnel light interference, the signal dynamic threshold is determined, and based on the processed signal set and the signal dynamic threshold, a reference point set is obtained; S2. Acquire reflection signal data of the UAV and the robot dog through the reference point set and extract reflection signal features; determine the initial position distribution of the UAV and the robot dog based on the reflection signal features; S3. Acquire dynamic reflection data of the UAV and the robot dog based on the initial position distribution, and process the dynamic reflection data to obtain a smooth position sequence; S3 includes the following steps: S31. Obtain the dynamic reflection data of the drone and the robot dog, and use a Kalman filter algorithm to fuse the time series to determine a position estimate; the drone or robot dog transmits an ultrasonic signal in the tunnel, which is received by a receiving device to form a time series, and the duration from signal transmission to signal return is recorded; S32. Determine dynamic change characteristics caused by movement based on the position estimate; and determine the continuity of the movement trajectory based on the dynamic change characteristics. S33. When the continuity of the moving trajectory meets the preset threshold, the position estimation value is optimized to obtain an updated position sequence; when the continuity of the moving trajectory does not meet the preset threshold, the process returns to S31; S34, calculating trajectory smoothness based on the updated position sequence, and adjusting parameters of a Kalman filter algorithm based on the trajectory smoothness, thereby obtaining the smoothed position sequence; S4. Calculating a relative motion trend based on the smoothed position sequence and the motion difference characteristics of the drone and the robot dog; and determining the position consistency of the drone and the robot dog based on the relative motion trend; The method for calculating the relative motion trend based on the smoothed position sequence and the motion difference characteristics of the drone and the robot dog comprises the following steps: S41, extracting motion trajectory points of the drone and the robot dog according to the smoothed position sequence, and determining motion characteristics of the drone and the robot dog respectively based on the motion trajectory points; S42, calculating the displacement difference between the drone and the robot dog at each time point based on the motion characteristics, thereby obtaining a velocity vector; S43, calculating a direction vector based on the angular change of the velocity vector to obtain the movement directions of the drone and the robot dog; and obtaining the relative movement trend based on the movement directions; The method for determining the position consistency of the drone and the robot dog based on the relative motion trend includes: S44, obtaining a preliminary position result set based on the acquired real-time reflection signal data and the relative motion trend; S45. Extracting reference coordinate data from the preliminary position result set to determine the relative position distribution of the drone and the robot dog; S46. Based on the relative position distribution, perform signal fusion processing on the real-time reflection signal to obtain an adjusted position solution result; determine the deviation between the adjusted position solution result and the relative motion trend; if the deviation exceeds a set deviation threshold, re-perform signal fusion processing; if the deviation is below the set deviation threshold, obtain updated position data of the drone and the robot dog, and thereby obtain the real-time position relationship between the drone and the robot dog; S47. Optimizing the motion trend based on the real-time position relationship to obtain an optimized position solution result; and determining the position consistency of the drone and the robot dog based on the optimized position solution result; Methods for determining the position consistency between the drone and the robot dog include: S471, extracting signal tracking deviation values ​​from the optimized position solution result to obtain a deviation value sequence; S472: Compare the values ​​in the deviation value sequence with a preset deviation threshold. If the deviation threshold is exceeded, determine the signal segment that needs to be resampled and obtain resampled data; otherwise, determine that the positions of the drone and the robot dog are consistent. S473, performing feature extraction on the resampled data to obtain a signal feature set; and performing smoothing processing on the signal features to obtain smoothed feature data; S474, calculating a position correction value based on the smoothing feature data to obtain position correction data; inputting the position correction data into S471 to extract a signal tracking deviation value; S5. Determine the real-time positioning coordinates of the drone and the robot dog based on the position consistency.

2. The method for joint positioning of a UAV and a robot dog in a tunnel based on prism reference points according to claim 1, characterized in that: S2 includes the following steps: S21. Acquire reflection signal data of the drone and the robot dog, perform feature extraction on the reflection signal data to obtain reflection signal features; and determine distribution characteristics of signal strength and signal distance based on the reflection signal features; S22. Calculate preliminary position estimates of the drone and the robot dog using triangulation based on the distribution characteristics of the signal strength and the signal distance; S23 , correcting the preliminary position estimate, calculating a matching degree between the corrected preliminary position estimate and the tunnel environment, and obtaining the initial position distribution.

3. A joint positioning system for a UAV and a robot dog in a tunnel based on a prism reference point, the system being used to implement the method according to any one of claims 1 to 2, characterized in that: include: Reference point acquisition module, feature extraction module, processing module, position calculation module and positioning module; The reference point acquisition module is used to construct a tunnel prism reference point array, obtain an initial reflection signal based on the prism reference point array, and pre-process the initial reflection signal to obtain a processed signal set; Combined with the distribution characteristics of tunnel light interference, the signal dynamic threshold is determined, and based on the processed signal set and the signal dynamic threshold, a reference point set is obtained; The feature extraction module obtains the reflected signal data of the UAV and the robot dog through the reference point set and extracts the reflected signal features; based on the reflected signal features, determines the initial position distribution of the UAV and the robot dog; The processing module is used to obtain dynamic reflection data of the UAV and the robot dog based on the initial position distribution, and process the dynamic reflection data to obtain a smooth position sequence; The position calculation module is used to calculate the relative motion trend based on the smooth position sequence and the motion difference characteristics of the drone and the robot dog; and determine the position consistency of the drone and the robot dog based on the relative motion trend; The positioning module is used to determine the real-time positioning coordinates of the drone and the robot dog based on the position consistency.

Images