An unmanned vehicle-based signal source progressive search and positioning method and device

Abstract

The application provides a signal source progressive search and positioning method and device based on an unmanned vehicle, relates to the field of unmanned vehicle autonomous navigation and signal source positioning, and gradually updates a counting rate data to accurately detect a source signal by filtering a collected counting rate through a sliding sampling window, updating a sampling window in real time through a FIFO principle to obtain an updated source signal strength value, preparing for calculating an optimal source search path; a global planner comprehensively evaluates a global path, a local planner plans an optimal path, and a target point is searched by the unmanned vehicle; after the target point is searched by the unmanned vehicle, a signal source detection limit value is set, a counting rate filtered after rotation is cyclically obtained, and then the signal source is positioned; the application gradually and in real time avoids obstacles and dynamically adjusts a search path, a path planning and path optimization closed-loop control process is formed through a termination judgment rule of a detection signal source limit value, and the autonomy and efficiency of the unmanned vehicle source positioning are improved.

Description

Technical Field

[0001] This application relates to the field of autonomous navigation and signal source localization for unmanned vehicles, specifically to a method and apparatus for progressive search and localization of signal sources based on unmanned vehicles. Background Technology

[0002] Nuclear, bio, and chemical technologies are widely used in medicine, energy, agriculture, and other fields, bringing enormous benefits to society, but also carrying potential risks. Once radioactive isotopes, highly toxic chemicals, or highly pathogenic pathogens are leaked due to human error, equipment malfunction, or deliberate sabotage, their rapid spread and wide-ranging impact can cause large-scale casualties, ecological disasters, and social panic in a short period. Quickly and accurately locating and pinpointing the source of pollution (i.e., source tracing) is the primary prerequisite for containing the spread of harm and ultimately eliminating the threat.

[0003] In recent years, unmanned and intelligent source-finding technologies, represented by driverless vehicles, drones, and robot swarms, have rapidly emerged and become an important direction for emergency response. Unmanned systems can replace or assist humans in entering high-risk areas, achieving rapid detection, data transmission, and autonomous positioning, significantly reducing the risk of personnel casualties.

[0004] However, existing source-finding methods mainly rely on detector counting to detect source-finding signals by directly superimposing environmental background effects. The detected source-finding signals are low, which can easily lead to misjudgment of the detected source-finding data. In the event of misjudgment of source-finding data, current source-finding methods separate the detection signal strength from the search and source-finding path planning, without fully considering the optimal path for obstacles and source-finding terrain, which limits the time cost of searching and locating the source-finding path. Summary of the Invention

[0005] 1) Technical problems to be solved Therefore, the present invention provides a progressive search and localization method and apparatus for signal sources based on unmanned vehicles, which solves the problems of misjudgment of source-finding data and the inability of the source-finding path to achieve the optimal time cost in existing source-finding methods.

[0006] (ii) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a method and apparatus for progressive search and localization of signal sources based on unmanned vehicles, comprising: The unmanned vehicle collects the count rate data generated by the signal source in real time through the signal detectors on board. The count rate is the number of times the detectors receive signals per unit time. The unmanned vehicle receives combined navigation data obtained from the ground base station in real time. The combined navigation data includes the position and attitude of the unmanned vehicle in the world coordinate system, so as to obtain the real-time coordinates and heading of the unmanned vehicle. A sliding sampling window is set up, the collected count rate is filtered, and the sampling window is updated according to the FIFO principle to obtain the filtered signal strength value; The unmanned vehicle rotates at its current position by a preset rotation angle, and the rotation difference is obtained by subtracting the count rate before rotation from the count rate after rotation. The azimuth angle between the signal source and the unmanned vehicle is determined by the rotation difference. Set the target point for the autonomous vehicle's movement, obtain the coordinates of the autonomous vehicle's current position, and calculate the coordinates of the autonomous vehicle's position when it reaches the target point; Based on the Navigation2 framework, the optimal path from the current location of the autonomous vehicle to the target point is found. This involves fusing the combined navigation data with synchronously acquired LiDAR point cloud data and visual sensor images to construct a grid map of the autonomous vehicle's path from its current location to the target point. Using the local planner within the Navigation2 framework, the obstacle avoidance path of the autonomous vehicle is calculated in real time. The optimal path is the one that minimizes the combined values ​​of following error, obstacle distance, and speed smoothness among different obstacle avoidance paths in the grid map. Speed ​​smoothness is a value that quantifies the stability of the autonomous vehicle's speed changes. After the unmanned vehicle finds the target point through the optimal path, it sets a limit value for signal source detection based on the minimum recognition signal value of the detector, and then obtains the filtered count rate before and after the unmanned vehicle rotates again. The count rate obtained again before and after the unmanned vehicle rotates is compared with the limit value to locate the signal source.

[0007] Furthermore, the sampling window and the fixed value window are set to filter the collected count rate. The sampling window is filtered by sliding the fixed value window. When the sampling window is not full, the arithmetic mean of the data in the sampling window is calculated. When the sampling window is full, the data in the sampling window is sorted, the maximum and minimum values ​​are removed, and the arithmetic mean of the remaining data in the sampling window is calculated. The arithmetic mean of the data within the sampling window is obtained. The earliest count rate data in the current sampling window is removed according to the FIFO principle. The data is then filtered through the sampling window to obtain the signal strength value of the filtered count rate.

[0008] Furthermore, the unmanned vehicle is rotated in place at its current position by the preset rotation angle. The filtered count rates before and after the rotation are obtained. The count rate after rotation is subtracted from the count rate before rotation to obtain the rotation difference. A signal difference threshold is preset for the rotation angle. If the rotation difference is greater than the signal difference threshold, the direction of the preset rotation angle is determined based on whether the signal difference threshold is greater than 0. The relationship between the rotation difference and the signal difference threshold is iteratively determined to make the signal difference threshold greater than the rotation difference, thereby determining the azimuth angle between the signal source and the unmanned vehicle.

[0009] Furthermore, the current position coordinates of the unmanned vehicle are obtained through the integrated navigation data, a forward travel amount is set, and the position coordinates of the unmanned vehicle reaching the target point are calculated based on the preset travel amount. The formula for the position coordinates of the unmanned vehicle reaching the target point is: ; in, The x-coordinate of the target point's location coordinates. The ordinate of the target point's position coordinates. The x-coordinate of the current position of the autonomous vehicle. The ordinate of the current position of the autonomous vehicle is y. The azimuth angle is... The travel distance is the stated travel amount.

[0010] Furthermore, the optimal path from the current position of the unmanned vehicle to the target point is found based on the Navigation2 framework structure. The Navigation2 framework includes a cost map layer unit. The cost map layer unit sets the grid resolution, fuses the LiDAR point cloud, the visual sensor image, and the combined navigation data to identify the global region from the current position of the unmanned vehicle to the target point, marks obstacle grids and idle grids, performs radius dilation processing on the obstacle grids to generate a dilated layer, and combines the pose information of the combined navigation data to mark the navigation cost layer of the drivable area, thereby constructing a grid map covering the unmanned vehicle from the current position to the target point.

[0011] Furthermore, the Navigation2 framework also includes a global planner and a local planner. The global planner obtains the current position coordinates of the autonomous vehicle and the position coordinates of the target point on the grid map, and plans a global path by comprehensively evaluating the actual cost and the estimated cost. The actual cost is the distance from the current position to any grid position in the grid map, and the estimated cost is the short distance from the current position to the target point. The global path is obtained by comparing the actual cost and the estimated cost. The local planner tracks the global path and calculates obstacle avoidance paths in real time during the autonomous vehicle's movement. The local planner enumerates different linear velocities and angular velocities to simulate and generate multiple grid node paths. It calculates the following error, obstacle distance, and speed smoothness of different node paths in the grid map, and finds the obstacle avoidance path with the smallest comprehensive value of the following error, obstacle distance, and speed smoothness as the optimal path, so that the autonomous vehicle can search for the target point according to the optimal path.

[0012] Furthermore, after the unmanned vehicle finds the target point through the optimal path, it sets a limit value for signal source detection based on the average value of the detector's background noise and the increment of the minimum recognition signal value. It then acquires the filtered count rate after the unmanned vehicle rotates again and determines whether the filtered count rate after the unmanned vehicle rotates is greater than the set limit value. If the filtered count rate after the unmanned vehicle rotates again is less than the set limit value, the unmanned vehicle continues to collect count rates to locate the signal source. If the filtered count rate after the unmanned vehicle rotates is greater than the set limit value, the location of the signal source is determined.

[0013] Furthermore, the data acquisition and perception module collects the count rate data output by the signal source in real time through the signal detectors on the unmanned vehicle, acquires the lidar point cloud data in real time through the lidar sensors in the unmanned vehicle, and acquires visual sensing images in real time through the binocular cameras on the unmanned vehicle. The navigation and positioning module allows the unmanned vehicle to receive combined navigation data from ground base stations in real time, obtain the coordinates of its current position through the ground base stations, and output the unmanned vehicle's position and attitude in the same coordinate system in real time; at the same time, it provides target points for dynamic navigation for path planning through the count rate data. The path planning module integrates the ROS 2 Navigation2 framework. The cost map layer unit fuses the LiDAR point cloud, the visual sensor image, and the combined navigation data to construct a raster map. The global planner generates a global path from the current location to the target point. The local planner uses the DWB algorithm to achieve real-time obstacle avoidance and dynamic adjustment of the signal source orientation. The chassis control module is used to receive the optimal speed command output by the path planning module and send it to the chassis driver through a communication protocol. The drive motor in the chassis driver controls the unmanned vehicle to achieve differential or omnidirectional movement. At the same time, according to the preset termination judgment rule, the position of the signal source is determined by the set signal difference threshold and the set limit value.

[0014] (iii) Beneficial effects: Compared with the prior art, this invention has the following beneficial effects: This invention provides a progressive search and localization method and apparatus for signal sources based on unmanned vehicles. By using a sliding sampling window to filter the collected count rate, and by updating the sampling window in real time according to the FIFO principle to obtain the updated source signal strength value, the progressively updated count rate data accurately detects the source signal, prepares for calculating the optimal source path, and enables the unmanned vehicle to approach the signal source in a short time, ensuring positioning accuracy and stability. The global planner comprehensively evaluates the global path, while the local planner enumerates different linear and angular velocities to simulate and generate multiple node path trajectories. The optimal obstacle avoidance path is found by minimizing the combined values ​​of node path following error, obstacle distance, and speed smoothness, enabling the autonomous vehicle to search for the target point. After the autonomous vehicle locates the target point via the obstacle avoidance path, a limit value for signal source detection is set, and the filtered count rate after rotation is obtained cyclically to pinpoint the signal source. This invention progressively and in real-time avoids obstacles and dynamically adjusts the search path through the global and local planners. Based on the termination judgment rule of the set signal source detection limit value, a closed-loop control process for path planning and optimization is formed, thereby achieving optimal time cost for the source-finding path and improving the autonomy and efficiency of the autonomous vehicle's source-finding and positioning. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall process of a progressive search and localization method for signal sources based on unmanned vehicles according to the present invention. Figure 2 An algorithm logic diagram of a progressive search and localization method for signal sources based on unmanned vehicles provided in an embodiment of the present invention; Figure 3 A working block diagram of a signal source progressive search and positioning device based on an unmanned vehicle provided in an embodiment of the present invention; In the diagram, 1 is the data acquisition and sensing module; 2 is the navigation and positioning module; 3 is the path planning module; and 4 is the chassis control module. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] like Figure 1 As shown, this invention provides a progressive search and localization method for signal sources based on an unmanned vehicle. The method involves: real-time acquisition of count rate data generated by the signal source using a signal detector mounted on the unmanned vehicle (the count rate being the number of signal events received by the detector per unit time); real-time acquisition of lidar point cloud data using a lidar sensor in the unmanned vehicle; real-time acquisition of visual sensing images using a binocular camera on the unmanned vehicle; and real-time reception of combined navigation data from a ground base station, including the unmanned vehicle's position and attitude in the world coordinate system, to obtain the unmanned vehicle's real-time coordinates and heading. Furthermore, a sliding sampling window is set up to filter the collected count rate. The sampling window is updated using the FIFO principle to obtain the filtered signal strength value. The unmanned vehicle rotates at its current position by a preset rotation angle. The filtered count rate before and after the rotation is obtained, and the rotation difference is calculated. The azimuth angle between the signal source and the unmanned vehicle is determined by the rotation difference. The target point for the unmanned vehicle is set, the position coordinates of the unmanned vehicle at its current position are obtained, the forward travel amount is set, and the position coordinates of the unmanned vehicle when it reaches the target point are calculated. Based on the Navigation2 framework, the optimal path from the current position to the target point of the autonomous vehicle is found. The LiDAR point cloud, visual sensor images and combined navigation data collected by the autonomous vehicle are integrated to construct a grid map from the current position to the target point. Based on the local planner in the Navigation2 framework, the obstacle avoidance path is calculated in real time during the autonomous vehicle's movement. The optimal obstacle avoidance path is the one with the smallest comprehensive value of following error, obstacle distance and speed smoothness under different paths in the grid map. After the unmanned vehicle finds the target point through the optimal path, it sets the limit value for signal source detection based on the detector's background noise and minimum recognition signal value. It then obtains the filtered count rate before and after the unmanned vehicle rotates again. The filtered count rate before and after the unmanned vehicle rotates is then compared with the set limit value for the signal source to locate the signal source.

[0018] like Figure 2 As shown, the main process of the progressive search and positioning method for unmanned vehicle signal sources in this invention is as follows: First, a signal detector is used to collect the count rate data generated by the signal source in real time, and the data is filtered through a sliding sampling window and a filtering window to obtain the filtered signal strength value. Then, a preset rotation angle is used to the side with a larger signal strength value, and the relationship between the rotation difference before and after the rotation and a preset threshold is determined. If the rotation difference before and after the rotation is greater than the preset threshold, the target point is calculated. Next, the relationship between the signal strength and the target limit is determined. If the signal strength is less than the target limit, the count rate data is collected again, and the relationship between the signal strength and the target limit is determined again. If the signal strength is greater than the target limit, an obstacle avoidance path is generated, and the target point is searched and located.

[0019] The signal detector in this invention includes a gamma-ray detector, One of the following: radio frequency detectors or electromagnetic wave energy detectors, including but not limited to the signal detectors mentioned above.

[0020] Furthermore, a sampling window and a fixed value window are set to filter the collected count rate. The sampling window is filtered by a sliding fixed value window. When the sampling window is not full, the arithmetic mean of the data in the sampling window is calculated. When the sampling window is full, the data in the sampling window is sorted, the maximum and minimum values ​​are removed, and the arithmetic mean of the remaining data in the sampling window is calculated. The arithmetic mean of the data within the sampling window is obtained. The earliest count rate data in the current sampling window is removed according to the FIFO principle. The data is then filtered through the sampling window in a loop to obtain the signal strength value of the filtered count rate.

[0021] In this embodiment, a length of [length] is established. The sliding sampling window with a length of Fixed value window, sliding sampling window length Fixed window length ,in When the sampling window is not full, the arithmetic mean of the data within the sampling window is calculated directly. When the sampling window is full, the data within the sampling window is sorted, the maximum and minimum values ​​are removed, and the remaining data are... The arithmetic mean of the data is calculated to obtain the filtered signal strength value. Whenever new data arrives, the window is updated according to the FIFO principle, the earliest acquired point in the current sampling window is removed, and then the arithmetic mean of the data in the sampling window is calculated repeatedly to obtain the filtered signal strength value.

[0022] By setting a sliding sampling window and a fixed value window, the counting rate collected by the unmanned vehicle at its current position is subjected to extreme value removal and mean filtering. Abnormal peaks caused by momentary occlusion or sensor jitter are eliminated, resulting in a smooth and reliable signal strength. This effectively suppresses counting statistical noise and ensures the robustness of target point orientation estimation.

[0023] Furthermore, the unmanned vehicle is rotated in place at its current position by a preset rotation angle. The count rate before and after the rotation is obtained after filtering, and the rotation difference is calculated. A signal difference threshold for the preset rotation angle is used. If the rotation difference is greater than the signal difference threshold, the direction of the preset rotation angle is determined based on whether the signal difference threshold is greater than 0. The relationship between the rotation difference and the signal difference threshold is iteratively judged to make the signal difference threshold greater than the rotation difference, thereby determining the azimuth angle between the signal source and the unmanned vehicle.

[0024] In this embodiment, the next forward direction of the autonomous vehicle is estimated, and the autonomous vehicle rotates in place by a preset angle N. The range is 30° to 90°, and the filtered count rate is obtained before and after rotation. Calculate the rotation difference : ; If the rotation difference is greater than the set negligible signal difference threshold, that is... According to The sign of the value determines the direction of the next rotation. The preset rotation angle is repeated, and the rotation difference is calculated until... Determine the azimuth angle between the signal source and the unmanned vehicle.

[0025] When the autonomous vehicle estimates its next direction of travel, it turns a small angle in place and then compares the signal strength before and after the rotation. It then compares whether the difference in signal strength during the rotation is greater than a set signal difference threshold. If so, it drives the autonomous vehicle to continue rotating in the direction of the stronger signal until the rotation difference is almost zero. At this point, it can be considered that the direction the vehicle is pointing is the location of the signal source. This progressive strategy of finding the target point avoids blind traversal and can approach the signal source in the shortest time. The unmanned operation avoids personnel entering high-radiation or dangerous areas.

[0026] Furthermore, the current position coordinates of the autonomous vehicle are obtained through integrated navigation data. A forward travel distance is set, and the position coordinates of the autonomous vehicle upon reaching the target point are calculated based on the preset travel distance. The formula for the position coordinates of the autonomous vehicle upon reaching the target point is: ; in, The x-coordinate of the target point's location is given. The ordinate of the target point's position is y. The x-coordinate of the current position of the autonomous vehicle is [x-coordinate]. The ordinate of the current position of the autonomous vehicle is y. It is the azimuth angle. This refers to the distance traveled.

[0027] In this embodiment, the next navigation target point for the autonomous vehicle is calculated using the vehicle's current coordinates. Using this as a baseline, and taking a distance of 2 meters on the horizontal plane as the travel distance, calculate the coordinates of the target point of the unmanned vehicle. for: ; Furthermore, based on the Navigation2 framework, the optimal path from the current position to the target point of the autonomous vehicle is found. The Navigation2 framework includes a cost map layer unit, which sets the grid resolution. It integrates LiDAR point cloud, visual sensor image and combined navigation data to identify the global area from the current position of the autonomous vehicle to the target point, marks obstacle grids and idle grids, performs radius dilation processing on obstacle grids to generate dilation layers, and combines the pose information of the combined navigation data to mark the navigation cost layer of the drivable area, thereby constructing a grid map covering the autonomous vehicle from the current position to the target point.

[0028] The Navigation2 framework also includes a global planner and a local planner. The global planner obtains the current position coordinates of the autonomous vehicle and the position coordinates of the target point on the grid map. It plans the global path by comprehensively evaluating the actual cost and the estimated cost. The actual cost is the distance from the current position to any grid position in the grid map, and the estimated cost is the short distance from the current position to the target point. The global path is obtained by comparing the actual cost and the estimated cost. The local planner tracks the global path and calculates obstacle avoidance paths in real time during the autonomous vehicle's movement. The local planner enumerates different linear and angular velocities to simulate and generate multiple grid node paths. It calculates the following error, obstacle distance, and speed smoothness of different node paths in the grid map, and finds the obstacle avoidance path with the smallest comprehensive value of following error, obstacle distance, and speed smoothness as the optimal path, so that the autonomous vehicle can search for the target point according to the optimal path.

[0029] In this embodiment, the unmanned vehicle is equipped with a source detector, a LiDAR device, and a binocular camera, as well as a combined navigation device capable of acquiring navigation data. The unmanned vehicle autonomously navigates to the target point. Based on the Navigation2 framework, it fuses LiDAR point clouds, visual sensor images, and combined navigation data to construct a grid map. It then calls the DWB local planner to generate an obstacle avoidance path and enumerates candidate linear velocities and angular velocities within robot dynamics constraints through actuators and mechanical systems. After simulating multiple short-range trajectories in the forward direction, the optimal path is selected based on a comprehensive consideration of path following error, obstacle distance, and speed smoothness. This enables real-time obstacle avoidance and global path tracking, allowing the autonomous vehicle to reach the target point. ; The Navigation2 framework consists of a cost map layer unit, a DWB local planner, and a global planner. The cost map layer unit is used to fuse LiDAR point clouds, visual sensor images, and integrated navigation data to construct a grid map. The DWB local planner plans the obstacle avoidance path and the global path that the autonomous vehicle follows while traveling in the grid map. The global planner generates the global path based on the path planned by the DWB local planner.

[0030] Global path planning uses a global planner to run the A* algorithm on a constructed grid map to generate an initial global path from the starting point to the ending point, and outputs a discrete path point sequence with the goal of minimizing the path following error.

[0031] Local trajectory optimization involves, after obtaining the global path, the DWB local planner enumerating candidate linear and angular velocities within the robot's dynamic constraints. The simulation combines multiple short-range trajectories, with each trajectory having a simulation duration of 2 seconds and a step size of 0.1 seconds. Then, based on a comprehensive cost function considering path following error, obstacle distance, and velocity smoothness, the trajectory with the lowest cost is selected. As the optimal control command, it enables real-time obstacle avoidance and global path tracking.

[0032] Obstacle avoidance is achieved through multi-sensor data fusion. During the journey, LiDAR point cloud, depth vision images, and color image data are fused in real time through a cost map layer. If an obstacle intrusion with an expansion radius of 0.3m is detected, local trajectory optimization is instantly re-executed to generate a new local trajectory. If the obstacle is dynamic, a predicted trajectory model is activated, and obstacles intersecting with the predicted trajectory are directly eliminated during the sampling phase. Candidates are selected to avoid the risk of collision.

[0033] By correcting the trajectory through a closed-loop integrated navigation system, the unmanned vehicle receives integrated navigation data in real time to achieve centimeter-level path tracking accuracy until the center of the vehicle reaches the target point within the tolerance radius, thus completing the path planning task.

[0034] Furthermore, after the unmanned vehicle finds the target point through the optimal path, it sets the limit value for signal source detection based on the detector's background noise and minimum recognition signal value. It then obtains the filtered count rate before and after the unmanned vehicle rotates again, and compares the obtained filtered count rate before and after the unmanned vehicle rotates with the set limit value for the signal source to locate the signal source.

[0035] Specifically, after the autonomous vehicle finds the target point through the optimal path, it sets a limit value for signal source detection based on the average value of the detector's background noise and the increment of the minimum recognition signal value. It then acquires the filtered count rate after the autonomous vehicle rotates again and determines whether the filtered count rate after the autonomous vehicle rotates is greater than the set limit value. If the filtered count rate after the autonomous vehicle rotates again is less than the set limit value, the system continues to collect the count rate through the autonomous vehicle to locate the signal source. If the acquired count rate after the autonomous vehicle rotates and the filtered count rate is greater than the set limit value, the location of the signal source is determined.

[0036] In this embodiment, after reaching the target point, a source-finding termination determination is performed, and the filtered count rate is obtained again. If the count rate is greater than the set signal source detection limit, then the condition is met. If the signal source is located, the search is stopped; otherwise, the search continues to iterate over the location of the signal source.

[0037] like Figure 3As shown, this invention discloses a progressive signal source search and positioning device based on an unmanned vehicle. The device includes a data acquisition and sensing module 1, a navigation and positioning module 2, a path planning module 3, and a chassis control module 4. The data acquisition and sensing module 1 collects the count rate data output by the signal source in real time through the signal detector mounted on the unmanned vehicle, and acquires the lidar point cloud data in real time through the lidar sensor in the unmanned vehicle. It also collects visual sensing images in real time through the binocular camera on the unmanned vehicle. The binocular camera and lidar on the unmanned vehicle collect dense depth maps, color images, and point cloud data, providing the system with high-precision three-dimensional environmental perception information.

[0038] The navigation and positioning module 2 receives combined navigation data from ground base stations in real time, obtains the coordinates of the current position through the ground base stations, and outputs the position and attitude of the unmanned vehicle in the same coordinate system in real time; at the same time, it provides target points for dynamic navigation for path planning through count rate data.

[0039] The path planning module 3 integrates the ROS 2 Navigation2 framework. The cost map layer unit fuses LiDAR point clouds, visual sensor images, and combined navigation data to construct a raster map. The global planner generates a global path from the current location to the target point. The local planner uses the DWB algorithm to achieve real-time obstacle avoidance and dynamic adjustment of the signal source orientation.

[0040] The chassis control module 4 receives the optimal speed command output by the path planning module 3 and sends it to the chassis driver through the communication protocol. The drive motor in the chassis driver controls the unmanned vehicle to achieve differential or omnidirectional movement. At the same time, according to the preset termination judgment rule, the position of the signal source is determined by the set signal difference threshold and the set limit value.

[0041] In this embodiment, the navigation and positioning module 2 obtains the current position coordinates of the unmanned vehicle through a ground RTK base station and outputs the position and attitude of the unmanned vehicle in the same coordinate system in real time. Simultaneously, it filters the signal data and calculates the azimuth of the signal source based on the filtered signal strength curve, converting the azimuth and distance into a forward target point in the world coordinate system, providing a dynamic navigation target point for path planning. This invention is not limited to obtaining the current position coordinates of the unmanned vehicle through a ground RTK base station; other methods can also be used to obtain the current position coordinates of the unmanned vehicle.

[0042] The chassis control module 4 receives the optimal speed command output by the path planning module 3 and sends it to the chassis driver via the CAN bus. The chassis driver drives the motor to achieve differential or omnidirectional movement. At the same time, according to the preset termination judgment rules, namely the signal strength threshold, the continuous stable convergence time threshold, and the distance threshold from the target point, it is determined whether the effective range of the signal source has been reached. Once any termination condition is met, braking and stopping are performed and a task completion signal is reported.

[0043] This invention provides a progressive search and localization method and apparatus for signal sources based on unmanned vehicles. By using a sliding sampling window, the collected count rate is filtered, and the sampling window is updated in real time according to the FIFO principle to obtain the updated source signal strength value. By progressively updating the count rate data, the source signal is accurately detected, preparing for the calculation of the optimal source path, enabling the unmanned vehicle to approach the signal source in a short time, and ensuring positioning accuracy and stability. The global planner comprehensively evaluates the global path, while the local planner enumerates different linear and angular velocities to simulate and generate multiple node path trajectories. The optimal obstacle avoidance path is found by minimizing the combined values ​​of node path following error, obstacle distance, and speed smoothness, enabling the autonomous vehicle to search for the target point. After the autonomous vehicle finds the target point through the obstacle avoidance path, a limit value for signal source detection is set, and the count rate after rotation and filtering is obtained cyclically to locate the signal source. This invention progressively avoids obstacles and dynamically adjusts the search path through the global and local planners. Based on the termination judgment rule of the set signal source detection limit value, a closed-loop control process of signal perception, path planning, obstacle avoidance execution, and path optimization is formed, thereby achieving the optimal time cost for the source-finding path. It can operate in complex terrains such as ruins, woodlands, and urban blocks, improving the autonomy and efficiency of the autonomous vehicle's source-finding and positioning.

[0044] It should be noted that the numerous details included in the above description are merely illustrative of the invention and not intended to limit it. In other embodiments of the invention, the method may have more, fewer, or different steps, and the order, inclusion, function, etc., of the steps may differ from those described and illustrated.

Claims

1. A progressive search and localization method for signal sources based on unmanned vehicles, characterized in that, The method includes the following steps: The unmanned vehicle collects the count rate data generated by the signal source in real time through the signal detectors on board. The count rate is the number of times the detectors receive signals per unit time. The unmanned vehicle receives combined navigation data obtained from the ground base station in real time. The combined navigation data includes the position and attitude of the unmanned vehicle in the world coordinate system, so as to obtain the real-time coordinates and heading of the unmanned vehicle. A sliding sampling window is set up, the collected count rate is filtered, and the sampling window is updated according to the FIFO principle to obtain the filtered signal strength value; The unmanned vehicle rotates at its current position by a preset rotation angle, and the rotation difference is obtained by subtracting the count rate before rotation from the count rate after rotation. The azimuth angle between the signal source and the unmanned vehicle is determined by the rotation difference. Set the target point for the autonomous vehicle's movement, obtain the coordinates of the autonomous vehicle's current position, and calculate the coordinates of the autonomous vehicle's position when it reaches the target point; Based on the Navigation2 framework, the optimal path from the current location of the autonomous vehicle to the target point is found. This involves fusing the combined navigation data with synchronously acquired LiDAR point cloud data and visual sensor images to construct a grid map of the autonomous vehicle's path from its current location to the target point. Using the local planner within the Navigation2 framework, the obstacle avoidance path of the autonomous vehicle is calculated in real time. The optimal path is the one that minimizes the combined values ​​of following error, obstacle distance, and speed smoothness among different obstacle avoidance paths in the grid map. Speed ​​smoothness is a value that quantifies the stability of the autonomous vehicle's speed changes. After the unmanned vehicle finds the target point through the optimal path, it sets a limit value for signal source detection based on the minimum recognition signal value of the detector, and then obtains the filtered count rate before and after the unmanned vehicle rotates again. The count rate obtained again before and after the unmanned vehicle rotates is compared with the limit value to locate the signal source.

2. The progressive search and localization method for signal sources based on unmanned vehicles according to claim 1, characterized in that, The sampling window and the fixed value window are set to filter the collected count rate. The sampling window is filtered by sliding the fixed value window. When the sampling window is not full, the arithmetic mean of the data in the sampling window is calculated. When the sampling window is full, the data in the sampling window is sorted, the maximum and minimum values ​​are removed, and the arithmetic mean of the remaining data in the sampling window is calculated. The arithmetic mean of the data within the sampling window is obtained. The earliest count rate data in the current sampling window is removed according to the FIFO principle. The data is then filtered through the sampling window to obtain the signal strength value of the filtered count rate.

3. The progressive search and localization method for signal sources based on unmanned vehicles according to claim 1, characterized in that, The unmanned vehicle is rotated in place at its current position by a preset rotation angle. The count rate before and after the rotation is obtained after filtering. The count rate after rotation is subtracted from the count rate before rotation to obtain the rotation difference. A signal difference threshold is preset for the rotation angle. If the rotation difference is greater than the signal difference threshold, the direction of the preset rotation angle is determined based on whether the signal difference threshold is greater than 0. The relationship between the rotation difference and the signal difference threshold is repeatedly judged to make the signal difference threshold greater than the rotation difference, thereby determining the azimuth angle between the signal source and the unmanned vehicle.

4. The progressive search and localization method for signal sources based on unmanned vehicles according to claim 1, characterized in that, The current position coordinates of the unmanned vehicle are obtained through the combined navigation data. A forward travel amount is set, and the position coordinates of the unmanned vehicle reaching the target point are calculated based on the preset travel amount. The formula for the position coordinates of the unmanned vehicle reaching the target point is: ； in, The x-coordinate of the target point's location coordinates. The ordinate of the target point's position coordinates. The x-coordinate of the current position of the autonomous vehicle. The ordinate of the current position of the autonomous vehicle is y. The azimuth angle is... The travel distance is the stated travel amount.

5. The progressive search and localization method for signal sources based on unmanned vehicles according to claim 1, characterized in that, The method for finding the optimal path from the current location of the unmanned vehicle to the target point based on the Navigation2 framework includes a cost map layer unit. The cost map layer unit sets the grid resolution, integrates the combined navigation data of the unmanned vehicle with the synchronously acquired LiDAR point cloud data and visual sensor images to identify the global region from the current location of the unmanned vehicle to the target point, marks obstacle grids and idle grids, performs radius dilation processing on obstacle grids to generate a dilated layer, and marks the navigation cost layer of the drivable area in combination with the pose information of the combined navigation data, thereby constructing a grid map covering the unmanned vehicle from the current location to the target point.

6. The progressive search and localization method for signal sources based on unmanned vehicles according to claim 5, characterized in that, The Navigation2 framework also includes a global planner and a local planner. The global planner obtains the current position coordinates of the autonomous vehicle and the position coordinates of the target point on the grid map, and plans a global path by comprehensively evaluating the actual cost and the estimated cost. The actual cost is the distance from the current position to any grid position in the grid map, and the estimated cost is the short distance from the current position to the target point. The global path is obtained by comparing the actual cost and the estimated cost. The local planner tracks the global path and calculates obstacle avoidance paths in real time during the autonomous vehicle's movement. The local planner enumerates different linear velocities and angular velocities to simulate and generate multiple grid node paths. It calculates the following error, obstacle distance, and speed smoothness of different node paths in the grid map, and finds the obstacle avoidance path with the smallest comprehensive value of the following error, obstacle distance, and speed smoothness as the optimal path, so that the autonomous vehicle can search for the target point according to the optimal path.

7. The progressive search and localization method for signal sources based on unmanned vehicles according to claim 6, characterized in that, After the unmanned vehicle finds the target point through the optimal path, it sets a limit value for signal source detection based on the average value of the detector's background noise and the increment of the minimum recognition signal value. It then acquires the filtered count rate after the unmanned vehicle rotates again and determines whether the filtered count rate after the unmanned vehicle rotates is greater than the set limit value. If the filtered count rate after the unmanned vehicle rotates again is less than the set limit value, the unmanned vehicle continues to collect count rates to locate the signal source. If the filtered count rate after the unmanned vehicle rotates is greater than the set limit value, the location of the signal source is located.

8. A progressive signal source search and positioning device based on an unmanned vehicle, characterized in that, The signal source progressive search and location device includes: The data acquisition and perception module collects the count rate data output by the signal source in real time through the signal detector on the unmanned vehicle, acquires the point cloud data of the lidar sensor in real time through the lidar sensor in the unmanned vehicle, and acquires visual sensing images in real time through the binocular camera on the unmanned vehicle. The navigation and positioning module allows the unmanned vehicle to receive combined navigation data from ground base stations in real time, obtain the coordinates of its current position through the ground base stations, and output the unmanned vehicle's position and attitude in the same coordinate system in real time; at the same time, it provides target points for dynamic navigation for path planning through the count rate data. The path planning module integrates the ROS 2 Navigation2 framework. The cost map layer unit fuses the LiDAR point cloud, the visual sensor image, and the combined navigation data to construct a raster map. The global planner generates a global path from the current location to the target point. The local planner uses the DWB algorithm to achieve real-time obstacle avoidance and dynamic adjustment of the signal source orientation. The chassis control module is used to receive the optimal speed command output by the path planning module and send it to the chassis driver through a communication protocol. The drive motor in the chassis driver controls the unmanned vehicle to achieve differential or omnidirectional movement. At the same time, according to the preset termination judgment rule, the position of the signal source is determined by the set signal difference threshold and the set limit value.

Images