Unmanned aerial vehicle full-autonomous navigation method, device and equipment for underground pipe gallery environment

Abstract

The invention provides an unmanned aerial vehicle full-autonomous navigation method, device and equipment for an underground pipe gallery environment, and relates to the technical field of unmanned aerial vehicle navigation. The method comprises the following steps: when full-autonomous navigation is carried out in an underground pipe gallery environment, firstly, a local grid map corresponding to the unmanned aerial vehicle in the underground pipe gallery environment at present can be constructed based on the current position of the unmanned aerial vehicle, the size of the unmanned aerial vehicle and the current flight path curvature radius of the unmanned aerial vehicle; determining whether the unmanned aerial vehicle is currently in a preset scene based on the current position of the unmanned aerial vehicle and the local grid map, wherein the preset scene comprises a height change scene or a single-side wall scene; and searching a lowest-cost path position based on a determination result, the current position of the unmanned aerial vehicle, the target position of the unmanned aerial vehicle and an improved artificial potential field method. By adopting the technical scheme provided by the invention, full-autonomous navigation of the unmanned aerial vehicle is realized, and particularly for narrow and dark unmanned aerial vehicle navigation scenes, the safety and efficiency of autonomous navigation of the unmanned aerial vehicle are effectively improved.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) navigation technology, and in particular to a fully autonomous navigation method, apparatus and equipment for UAVs in underground utility tunnel environments. Background Technology

[0002] Underground utility tunnels are public tunnels located underground in cities, used for the centralized laying of municipal pipelines such as electricity, communications, and gas. Inspecting underground utility tunnels is a complex and dangerous task, mainly due to the high intensity of work, long work cycles, and harsh environments along some routes. Traditional manual inspection methods face significant challenges. Therefore, the use of drones can be considered to accomplish the inspection of underground utility tunnels.

[0003] To accomplish the inspection of underground utility tunnels, the primary challenge is addressing the navigation problem of unmanned aerial vehicles (UAVs). However, given the narrow, dark environment of underground utility tunnels and the lack of environmental information, UAV navigation faces new challenges. Therefore, achieving fully autonomous navigation for UAVs in underground utility tunnel environments is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] This application provides a fully autonomous navigation method, apparatus, and device for UAVs in underground utility tunnel environments, which realizes fully autonomous navigation of UAVs in underground utility tunnel environments, especially for narrow and dark UAV navigation scenarios, effectively improving the safety and efficiency of UAV autonomous navigation.

[0005] This application provides a fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments, including: Based on the drone's current location, the drone's size, and the radius of curvature of the drone's current flight trajectory, a local grid map corresponding to the drone's current location in the underground utility tunnel environment is constructed. Based on the current position of the drone and the local grid map, it is determined whether the drone is currently in a preset scene, which includes a scene with changes in altitude or a scene with a single wall. Based on the determined results, the current position of the UAV, the target position of the UAV, and the improved artificial potential field method, the lowest cost path position is searched. Once the location of the lowest-cost path is determined, fully autonomous navigation is performed based on that location.

[0006] According to the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments provided in this application, the step of searching for the lowest-cost path based on the determined results, the target position of the UAV, and an improved artificial potential field method includes: If the determination result indicates that the UAV is not currently in the preset scenario, the lowest cost path location is searched based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method under the current flight parameters of the UAV. or, If the determination result indicates that the UAV is currently in a preset scenario, the current flight parameters of the UAV are corrected, and under the corrected flight parameters, the lowest cost path position is searched based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method.

[0007] According to the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments provided in this application, the step of correcting the current flight parameters of the UAV when the determination result indicates that the UAV is currently in a preset scenario includes: When the drone is currently in a scenario of altitude change, the drone's current flight altitude is corrected based on the width of the vertical region of the drone's current flight path. or, When the drone is currently in a scenario with a single wall, the corresponding straight line model of the wall is determined based on the RANSAC algorithm, and the current main heading of the drone is corrected based on the straight line model of the wall.

[0008] According to the fully autonomous navigation method for underground utility tunnel environments provided in this application, the step of determining the corresponding straight wall model based on the RANSAC algorithm includes: Based on the current main heading of the UAV and the local grid map, obtain a set of wall edge points for navigation; For the set of wall edge points, the initial straight-line model wall with the most interior points is determined based on the RANSAC algorithm; The wall straight model is determined based on the vertical distance between each interior point and the initial straight model wall.

[0009] According to the fully autonomous navigation method for underground utility tunnel environments provided in this application, the step of determining the straight-line model of the wall based on the vertical distance between each interior point and the initial straight-line model wall includes: For each interior point, the weight corresponding to the interior point is determined based on the vertical distance between the interior point and the wall of the initial straight line model; Based on the vertical distance and weight of each interior point, a target loss function is constructed. The target loss function is used to characterize the degree of matching between all interior points and the initial straight line model wall. The target loss function is solved using the weighted least squares method until the number of iterations reaches a preset threshold, thus obtaining the straight line model of the wall.

[0010] According to the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments provided in this application, the step of searching for the lowest-cost path location based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method includes: Based on the current position and target position of the UAV, the repulsive and attractive forces corresponding to the current position are calculated using the improved artificial potential field method. Based on the repulsive and attractive forces, a target navigation area is determined from the local grid map, the target navigation area including multiple grids; For each grid, the passage cost value corresponding to the center point of the grid is determined, wherein the smaller the passage cost value, the farther the center point is from the obstacle; The center point of the grid corresponding to the minimum passage cost value among the passage cost values ​​of the center point of each grid is determined as the location of the lowest cost path.

[0011] This application also provides a fully autonomous navigation device for unmanned aerial vehicles (UAVs) in underground utility tunnel environments, including: The construction unit is used to construct a local grid map of the UAV in the current environment facing the underground utility tunnel, based on the UAV's current position, the UAV's size, and the radius of curvature of the UAV's current flight trajectory. The determining unit is used to determine whether the drone is currently in a preset scene based on the drone's current position and the local grid map. The preset scene includes a scene with changes in altitude or a scene with a single wall. The search unit is used to search for the lowest cost path location based on the determined results, the current position of the UAV, the target position of the UAV, and the improved artificial potential field method. The navigation unit is used to perform fully autonomous navigation based on the lowest-cost path location once the lowest-cost path location has been found.

[0012] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments as described in any of the preceding claims.

[0013] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments as described in any of the preceding claims.

[0014] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments as described in any of the preceding claims.

[0015] The fully autonomous navigation method, apparatus, and equipment for UAVs in underground utility tunnel environments provided in this application, when performing fully autonomous navigation in such environments, first construct a local grid map corresponding to the UAV's current location, size, and current flight trajectory curvature radius based on the UAV's current position. Then, based on the UAV's current position and the local grid map, it determines whether the UAV is currently in a preset scenario, including scenarios with altitude changes or single-sided walls. Finally, based on the determined results, the UAV's current position, target position, and an improved artificial potential field method, it searches for the lowest-cost path position. This method, which determines whether the UAV is in a preset scenario based on its current position and the local grid map, and searches for the lowest-cost path position using the determined results, the UAV's current position, target position, and the improved artificial potential field method, enables autonomous flight and obstacle avoidance, achieving fully autonomous navigation for the UAV. This is particularly beneficial for narrow and dark UAV navigation scenarios, effectively improving the safety and efficiency of autonomous UAV navigation. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments, as provided in an embodiment of this application.

[0018] Figure 2 This is a schematic diagram of a process for determining a straight line model of a wall, provided in an embodiment of this application.

[0019] Figure 3 This is a schematic diagram illustrating a process for searching the lowest-cost path location based on the current location of the UAV, the target location of the UAV, and an improved artificial potential field method, as provided in an embodiment of this application.

[0020] Figure 4 This is a structural schematic diagram of a fully autonomous navigation device for underground utility tunnel environments, provided as an embodiment of this application.

[0021] Figure 5This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] In the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone, where A and B can be singular or plural. In the textual description of this application, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0024] The technical solution provided in this application can be applied to harsh environments, such as narrow and dark drone navigation scenarios. Taking underground utility tunnels as an example, underground utility tunnel inspection is a relatively tedious and dangerous task, mainly facing problems such as high work intensity, long work cycle, and harsh environment of some lines. Traditional manual inspection methods face huge challenges. Therefore, it is possible to consider using drones to realize the inspection task of underground utility tunnels.

[0025] To accomplish the inspection of underground utility tunnels, the primary challenge is addressing the navigation problem of unmanned aerial vehicles (UAVs). However, given the narrow, dark environment of underground utility tunnels and the lack of environmental information, UAV navigation faces new challenges. Therefore, achieving fully autonomous navigation for UAVs in underground utility tunnel environments is a technical problem that urgently needs to be solved by those skilled in the art.

[0026] To achieve fully autonomous navigation of unmanned aerial vehicles (UAVs) in underground utility tunnel environments, this application provides a method for such navigation. First, based on the UAV's current position, size, and current flight trajectory curvature radius, a local grid map is constructed corresponding to the UAV's current location within the underground utility tunnel environment. Then, based on the UAV's current position and the local grid map, it is determined whether the UAV is in a preset scenario, including scenarios with altitude changes or single-sided walls. Next, based on the determined results, the UAV's current position, target position, and an improved artificial potential field method, a minimum-cost path is searched. This method, which determines whether the UAV is in a preset scenario based on its current position and the local grid map, and searches for the minimum-cost path based on the determined results, the UAV's current position, target position, and the improved artificial potential field method, enables autonomous flight and obstacle avoidance, achieving fully autonomous navigation. This is particularly beneficial for narrow and dark UAV navigation scenarios, effectively improving the safety and efficiency of autonomous UAV navigation.

[0027] The fully autonomous navigation method for underground utility tunnel environments provided in this application will be described in detail below through several specific embodiments. It is understood that these specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0028] Figure 1 This application provides a flowchart illustrating a fully autonomous navigation method for unmanned aerial vehicles (UAVs) in an underground utility tunnel environment. This method can be applied to UAVs, and the executing entity can be either the UAV itself or a fully autonomous navigation device for underground utility tunnel environments installed within it. This fully autonomous navigation device can be implemented through software, hardware, or a combination of both, and can be configured according to actual needs. For example, please refer to... Figure 1 As shown, this fully autonomous navigation method for UAVs in underground utility tunnel environments may include: S101. Based on the current location of the UAV, the size of the UAV, and the radius of curvature of the current flight trajectory of the UAV, construct a local grid map corresponding to the current UAV in the environment facing the underground utility tunnel.

[0029] For example, when constructing a local grid map of the UAV facing an underground utility tunnel environment based on the UAV's current position, size, and current flight trajectory curvature radius, the UAV's main heading can be initialized first, with the initial main heading being the direction of the utility tunnel. The current main heading of the UAV can then be initialized as the main heading. A global two-dimensional grid map can be constructed using the Cartographer algorithm. The size of the local grid map can be determined based on the UAV's current position, size, and current flight trajectory curvature radius. Thus, the local grid map can be extracted from the global two-dimensional grid map based on the size of the local grid map.

[0030] For example, when determining the size of a local grid map based on the drone's current position, the drone's size, and the radius of curvature of the drone's current flight trajectory, the drone's current position can be selected as the center point of the local grid map. And according to the size of the drone With the radius of curvature of the flight trajectory Determine the size of the required local sliding window raster map. Due to the radius of curvature of the flight trajectory This may change over time, assuming the drone's left and right depth is... and Then the radius of curvature of the flight trajectory The size of the corresponding local raster map ,in, This represents a given threshold, which can be set according to actual needs.

[0031] In scenarios where a global 2D grid map already exists, such as one obtained using a 2D SLAM algorithm like Cartographer, a local grid map is extracted from the global map according to the specified size. In scenarios without a prior global map, a sliding window approach is used to directly maintain a local grid map within the sensor's field of view to support real-time navigation.

[0032] S102. Determine whether the drone is currently in a preset scene based on the drone's current location and local grid map. The preset scenes include scenes with changes in altitude or scenes with a single wall.

[0033] Among them, altitude change scenarios refer to scenarios where the terrain altitude directly in front of or below the drone changes significantly during horizontal flight. For example, the drone enters an area with an incline, such as a ramp in a utility tunnel, a stairwell, or an outdoor environment with undulating terrain.

[0034] A single-wall scenario refers to a situation where a drone is flying very close to a large, continuous obstacle, i.e., a wall, on one side, while the other side is a relatively open space. For example, a drone may be flying along one side wall of a utility tunnel, the exterior wall of a building, or the pillars of a corridor.

[0035] For example, when determining whether a drone is in a scene of altitude change based on its current position and a local grid map, a gradient calculation method can be used. First, based on the drone's current position point cloud coordinates, its forward point cloud is divided into an upper point cloud set and a lower point cloud set, which can be denoted as follows: and And set a constant threshold. If satisfied If the drone is in a changing altitude scenario, it can be determined that the drone is currently in a changing altitude scenario. Alternatively, other methods can be used, such as plane fitting, to fit a plane to the grid points in front of the drone. If the angle between the fitted plane normal vector and the horizontal plane normal vector is too large, for example, exceeding 15 degrees, it indicates that the current situation is a slope, and the drone is currently in a changing altitude scenario.

[0036] For example, to determine whether a drone is currently in a single-wall scenario based on its current position and a local grid map, obstacle density analysis can be used. This involves dividing the grid map around the drone into multiple sectors by angle, such as two 180-degree sectors to the left and right; and calculating the number or density of obstacle grids in each sector. If the obstacle density on one side is significantly higher than on the other side, for example, more than three times higher, then the drone is determined to be in a single-wall scenario. Alternatively, other methods can be used, such as feature extraction methods. Algorithms like RANSAC can be used to fit a straight line to the point cloud data around the drone. If a long straight line representing a wall can be stably fitted on one side, while the point cloud on the other side is sparse, then the drone is determined to be in a single-wall scenario.

[0037] S103. Based on the determined results, the current position of the UAV, the target position of the UAV, and the improved artificial potential field method, search for the lowest cost path position.

[0038] The lowest-cost path position can be understood as the drone's position at the next moment. Thus, after controlling the drone to travel from its current position to the lowest-cost path position, the next iteration is performed. That is, the lowest-cost path position is redefined as the drone's current position, and the above-mentioned S101-S103 related operations are repeated to determine a new lowest-cost path position. This allows the drone to travel according to the determined lowest-cost path position at each moment until the drone reaches the target position, realizing fully autonomous navigation of the drone and effectively improving the safety and efficiency of the drone's autonomous navigation.

[0039] If the location of the lowest-cost path is not found, wait in place and fine-tune the main course; if the location of the lowest-cost path is found, execute the following S104: S104. If the location of the lowest cost path is found, perform fully autonomous navigation based on the location of the lowest cost path.

[0040] As can be seen from the embodiments of this application, when performing fully autonomous navigation in an underground utility tunnel environment, a local grid map corresponding to the UAV's current location in the underground utility tunnel environment can be constructed first based on the UAV's current position, UAV size, and the radius of curvature of the UAV's current flight trajectory. Then, based on the UAV's current position and the local grid map, it can be determined whether the UAV is currently in a preset scenario, including scenarios with altitude changes or single-sided walls. Finally, based on the determination results, the UAV's current position, the UAV's target position, and the improved artificial potential field method, the lowest-cost path position is searched. In this way, by determining whether the UAV is currently in a preset scenario based on its current position and the local grid map, and by jointly searching for the lowest-cost path position based on the determination results, the UAV's current position, the UAV's target position, and the improved artificial potential field method, autonomous flight and obstacle avoidance are achieved based on the lowest-cost path position, realizing fully autonomous navigation of the UAV. This is especially beneficial for narrow and dark UAV navigation scenarios, effectively improving the safety and efficiency of autonomous UAV navigation.

[0041] Based on the above Figure 1 The illustrated embodiment, for example, in S103 above, when searching for the lowest-cost path location based on the determined result, the target position of the UAV, and the improved artificial potential field method, can include at least two of the following possible scenarios: In one possible scenario, the determination result indicates that the drone is not currently in the preset scenario.

[0042] If the result indicates that the UAV is not currently in the preset scenario, then, under the current flight parameters of the UAV, the lowest cost path location is searched based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method.

[0043] In another possible scenario, the determination result indicates that the drone is currently in a preset scenario.

[0044] If the result indicates that the drone is currently in a preset scenario, the drone's current flight parameters can be corrected first. Then, based on the corrected flight parameters, the lowest cost path position can be searched using the drone's current position, the drone's target position, and the improved artificial potential field method.

[0045] For example, in the embodiments of this application, when the determination result indicates that the drone is currently in a preset scenario, the correction of the drone's current flight parameters may include the following two cases: Scenario 1: When the drone is currently in a scenario of altitude change, adjust the drone's current flight altitude based on the width of the upper and lower regions of the drone's current flight path.

[0046] For example, when adjusting the drone's current flight altitude based on the width of the vertical region of its current flight path, if the width below narrows (e.g., uphill), the drone's altitude is increased to maintain a safe distance from the ground; if the width above narrows (e.g., encountering a crossbeam), the drone's altitude is decreased; if both vertical and vertical space are limited, a safe, intermediate altitude is selected. This dynamic adjustment of the drone's flight altitude ensures that the drone always has maximum safe maneuverability in complex vertical structures.

[0047] Scenario 2: When the drone is currently in a scenario with a single wall, the corresponding straight line model of the wall is determined based on the RANSAC algorithm, and the drone's current main heading is corrected based on the straight line model of the wall.

[0048] For example, the initial main heading of the UAV should be the direction of the pipe gallery. The initial main heading of the UAV is the main heading. To ensure that the main heading of the UAV does not deviate from the pipe, the UAV will determine the corresponding straight line model of the wall in real time based on the RANSAC algorithm during the flight. The UAV’s current main heading is then corrected based on the straight line model of the wall, optimizing the UAV’s fully autonomous navigation path. Moreover, it can quickly and accurately obtain obstacle and wall information in complex scenarios (such as single-sided walls and terrain changes) to achieve robust positioning.

[0049] The RANSAC algorithm is a mathematical model estimation method that randomly selects the minimum number of samples that satisfy the model from the model data. It uses the model's remainder set to test the randomly selected samples, and then iteratively searches for the optimal model. For an example, see [link to example]. Figure 2 As shown, Figure 2 This application provides a flowchart illustrating the process of determining a straight line model of a wall, which may include: S201. Based on the UAV's current main heading and local grid map, obtain a set of wall edge points for navigation.

[0050] For example, when acquiring the set of wall edge points for navigation based on the UAV's current main heading and local grid map, a forward-extending fan-shaped or rectangular region of interest can be determined using the UAV's current main heading and local grid map as the search area for potential navigation walls. The point cloud data collected by environmental perception sensors (such as LiDAR or depth cameras) within this search area is then filtered to remove noise points that are too close or too far away. Next, based on the continuity or density clustering of the point cloud, the subset of point clouds most likely belonging to linear walls is extracted. Finally, this spatially filtered and preprocessed subset of point clouds is used as the set of wall edge points for navigation. This method of acquiring the set of wall edge points for navigation significantly reduces the amount of data and outlier interference, thereby significantly improving the efficiency of acquiring the straight-line wall model.

[0051] S202. For the set of wall edge points, determine the initial straight line model wall with the most interior points based on the RANSAC algorithm.

[0052] For example, in an embodiment of this application, when determining the initial straight-line model wall with the most interior points based on the RANSAC algorithm, it may include: A. Randomly select the minimum number of points needed to determine the equation of the straight line from the set of points on the wall edge, i.e., 2 target points. Let the gradient direction of target point 1 be... The gradient direction of target point 2 is If two randomly selected target points lie on the same straight line, the absolute value of the difference in gradient directions between the two target points should be within a certain range. A custom threshold for the difference in gradient directions between the two target points can be defined. See Formula 1 below: Formula 1 If the gradient directions of the two extracted target points do not satisfy Formula 1 above, then target points are re-extracted; if the gradient directions of the two extracted target points satisfy Formula 1 above, then the initial straight line model wall is calculated using these two target points, and the average gradient direction of these two target points is taken as the main direction of the gradient direction of the straight line. .

[0053] b. Traverse the remaining edge points in the wall edge point set, excluding the two target points mentioned above. If the currently traversed edge point is a point on a straight line, then the edge point should satisfy the following condition: the absolute value of the interpolation between the gradient direction of the edge point and the principal direction of the gradient direction of the straight line should be less than a threshold. See Formula 2 below; and the distance from the currently traversed edge point to the line is less than the distance threshold. If two conditions are met, the currently traversed edge point is determined to be an interior point; otherwise, the currently traversed edge point is determined to be an exterior point, i.e., a noise point.

[0054] Formula 2 in, This represents the gradient direction of the currently traversed edge point, and the distance from the currently traversed edge point to the line. See Formula 3 below: Formula 3 c. After the traversal is complete, calculate the proportion of the interior points. =Number of interior points / Total number of edge points included in the wall edge point set.

[0055] d. Repeat step ac above for iteration until the number of iterations reaches the iteration termination number. This yields the initial straight-line model wall with the most interior points. The iteration termination number is [number missing]. The calculation can be performed theoretically, and the formula is shown in Formula 4 below: Formula 4 in, This represents the probability of drawing a good sample from the sample, and is generally set to... Within the range, This represents the minimum number of points required for wall estimation using a linear model. This indicates the proportion of interior points in the sample points.

[0056] Typically, at a confidence level of Under these conditions, after multiple iterations, there should be at least one sampling that results in the selected... Each sample set contains interior points, thus obtaining the initial straight-line model wall with the most interior points determined based on the RANSAC algorithm.

[0057] After determining the initial straight-line model wall with the most interior points based on the RANSAC algorithm, in order to further prevent misfitting of the straight line and obtain more accurate straight-line parameters, for example, in the embodiments of this application, a more accurate wall straight-line model can be determined based on the vertical distance between each interior point and the initial straight-line model wall, that is, the following S203 is executed: S203. Determine the straight wall model based on the vertical distance between each internal point and the initial straight wall model.

[0058] For example, in this embodiment of the application, when determining the straight wall model based on the vertical distance between each interior point and the wall of the initial straight wall model, for each interior point, the weight corresponding to the interior point can be determined first based on the vertical distance between the interior point and the wall of the initial straight wall model; based on the vertical distance and weight corresponding to each interior point, a target loss function is constructed, which is used to characterize the degree of matching between all interior points and the wall of the initial straight wall model; then the target loss function is solved based on the weighted least squares method until the number of iterations reaches a preset threshold, and the straight wall model is obtained. In this way, the current main heading of the UAV can be corrected based on the straight wall model.

[0059] For example, when determining the weight of an interior point based on its vertical distance from the initial straight-line model wall, the weight of the interior point is determined by the first... Taking an internal point as an example , num The slope of the wall in the initial straight-line model can be denoted as the total quantity of the interior. The intercept can be denoted as , No. The corresponding internal point can be denoted as: The weights corresponding to the interior points can be seen in Formula 5 below: Formula 5 in, This represents the vertical distance between an interior point and the wall of the initial straight-line model, and can be used as a weighted function variable. This represents the distance threshold.

[0060] When constructing the target loss function based on the vertical distance and weight of each interior point, the vertical distance from each interior point to the wall of the initial straight line model can be weighted. For example, the target loss function can be seen in the following formula 6: Formula 6 in, Indicates the position coordinates of an interior point.

[0061] The target loss function is solved by weighted least squares until the number of iterations reaches a preset threshold to obtain an accurate straight-line model of the wall. This allows the UAV's current main heading to be corrected based on the straight-line model of the wall.

[0062] For example, in an embodiment of this application, when correcting the current main heading of the UAV based on the straight line model of the wall, see Formula 7 below: Formula 7 in, This indicates the modified main heading of the drone. Indicates the drone's current main heading. This indicates that the slope of the left wall was detected based on the RANSAC algorithm. This indicates that the slope of the right wall was detected based on the RANSAC algorithm. This indicates that the wall type was detected based on the RANSAC algorithm.

[0063] For example, in the two possible scenarios described above, the specific implementation of searching for the lowest-cost path location based on the UAV's current position, the UAV's target position, and the improved artificial potential field method can be found below. Figure 3 The example shown.

[0064] Figure 3 This application provides a flowchart illustrating a process for searching the lowest-cost path location based on the current location of the UAV, the target location of the UAV, and an improved artificial potential field method, as illustrated in the embodiments of this application. For example, please refer to [link to relevant documentation]. Figure 3 As shown, the method may include: S301. Based on the current position and target position of the UAV, calculate the repulsive and attractive forces corresponding to the current position using the improved artificial potential field method.

[0065] Under normal circumstances, when calculating the repulsive force corresponding to the current position, a potential field is established in the neighborhood of the drone's current position. Obstacles will generate repulsive forces, while safe areas will generate attractive forces. The drone will move along the direction of the resultant force of the repulsive and attractive forces. The repulsive potential field function... See Formula 8 below: Formula 8 In formula 8 above, This represents the proportionality coefficient. This represents a vector, with a size equal to the drone's current position. relative to the location of obstacles The Euclidean distance between them, with the direction from the obstacle towards the drone, is a constant, representing the maximum distance at which an obstacle can affect the drone.

[0066] However, in this embodiment, considering that when the drone encounters an obstacle, the repulsive force generated will increase as the distance between the drone and the obstacle decreases, and when the distance between the drone and the obstacle approaches 0, the repulsive force will approach infinity, in order to reduce drone shaking, in this embodiment, the repulsive potential field function can be reconstructed, as shown in Equation 9 below: Formula 9 Based on the repulsive potential field function shown in Formula 9 above, the corresponding formula for calculating the repulsive field can be found in Formula 10 below: Formula 10 In Formula 10, Indicates the coefficient of repulsive force. This indicates the adjustment factor for the range of repulsive force. This represents an adjustment factor related to the distance between the drone and the obstacle. Mathematically, it can be intuitively seen that as the distance between the drone and the obstacle continuously decreases, the repulsive force will eventually tend to a constant value, thus allowing the calculation of the repulsive force corresponding to the current position.

[0067] For example, using the current position and target position of the UAV as a reference, and based on the improved artificial potential field method, the gravitational force corresponding to the current position can be calculated. This can be achieved by constructing a gravitational potential field function, including: first calculating the distance between the current position and the target position of the UAV, and then applying the gravitational field formula... Generates gravitational potential energy; among which, Represents the gravitational gain coefficient. This represents the distance between the drone and the obstacle; then, by calculating the negative gradient with respect to the gravitational field, we obtain the gravitational vector. , Indicates the target location. This indicates the drone's current position. By dynamically adjusting the gravity gain coefficient (e.g., the gravity increases with distance) or optimizing the gravity direction using a local grid map, the drone can be ensured to stably approach the target even in complex environments.

[0068] S302. Based on repulsion and attraction, determine the target navigation area from the local grid map. The target navigation area includes multiple grids.

[0069] For example, when determining the target navigation area from a local grid map based on repulsive and attractive forces, the repulsive and attractive forces can be vector-superimposed at the center point of each grid cell to obtain a resultant force vector field. Then, starting from the grid cell at the UAV's current position, along the approximate direction pointing to the target in the resultant force vector field, a continuous grid region is selected as a candidate region within a preset fan-shaped or rectangular search range. The criteria for determining this candidate region are: the resultant force direction of all grid cells within the region should have high consistency, and the resultant force magnitude should be moderate, neither too strong (leading to approach obstacles) nor too weak (leading to deviate from the target). In this way, a continuous set of grid cells that meets the requirements of directional consistency and safety can be determined as the target navigation area.

[0070] S303. For each grid, determine the passage cost value corresponding to the center point of the grid, wherein the smaller the passage cost value, the farther the center point is from the obstacle.

[0071] The passage cost value corresponding to the center point of the grid is negatively correlated with the distance from the center point to the nearest impassable area.

[0072] To determine the first Taking the passage cost value corresponding to the center point of a grid as an example, see Formula 11 below: Formula 11 in, This indicates that the drone is in the The passage cost value of the center point of each grid. Indicates the first The coordinates of the center point of each grid cell Indicates the first element in a local raster map The spatial location of obstacles near each grid.

[0073] S304. Among the passage cost values ​​corresponding to the center points of each grid, the center point of the grid corresponding to the minimum passage cost value is determined as the location of the lowest cost path.

[0074] As can be seen, in this embodiment, when searching for the lowest-cost path position based on the UAV's current position, the UAV's target position, and the improved artificial potential field method, the repulsive and attractive forces corresponding to the current position are first calculated based on the improved artificial potential field method. Then, based on the repulsive and attractive forces, the target navigation area is determined from the local grid map. This allows for the initial determination of a rough target navigation area based on the repulsive and attractive forces, effectively preventing the UAV from mistakenly entering a "trap" with extremely high repulsive forces (i.e., close to obstacles) due to simply chasing the lowest-cost point. This macroscopically avoids high-risk areas and greatly enhances the robustness of obstacle avoidance. Based on this, the passage cost value corresponding to the center point of each grid is then combined to determine the target navigation area. By defining the lowest-cost path position, only the passage cost value corresponding to the center point of a small number of grids needs to be calculated to finally determine the lowest-cost path position. This divide-and-conquer strategy satisfies both local optima (minimum cost) and global trends (within the target navigation area). This makes the generated path no longer the oscillating or jagged path that may occur in the traditional artificial potential field method, but a smoother, continuous, and high-quality path that directly points to the target. This allows the UAV to achieve autonomous flight and obstacle avoidance based on the lowest-cost path position, realizing fully autonomous navigation of the UAV. Especially for narrow and dark UAV navigation scenarios, it effectively improves the safety and efficiency of UAV autonomous navigation.

[0075] The fully autonomous navigation device for UAVs in underground utility tunnel environments provided in this application is described below. The fully autonomous navigation device for UAVs in underground utility tunnel environments described below can be referred to in correspondence with the fully autonomous navigation method for UAVs in underground utility tunnel environments described above.

[0076] Figure 4 A schematic diagram of a fully autonomous navigation device for an underground utility tunnel environment is provided as an embodiment of this application. For example, please refer to [link to relevant documentation]. Figure 4 As shown, the fully autonomous navigation device 40 for underground utility tunnel environments may include: Construction unit 401 is used to construct a local grid map of the UAV in the current environment facing the underground utility tunnel based on the current position of the UAV, the size of the UAV, and the radius of curvature of the current flight trajectory of the UAV. The determining unit 402 is used to determine whether the drone is currently in a preset scene based on the drone's current position and the local grid map. The preset scene includes a scene with changes in altitude or a scene with a single wall. Search unit 403 is used to search for the lowest cost path location based on the determined result, the current position of the UAV, the target position of the UAV, and the improved artificial potential field method. The navigation unit 404 is used to perform fully autonomous navigation based on the lowest cost path location when the lowest cost path location is found.

[0077] For example, in an embodiment of this application, the search unit 403 is used to search for the lowest-cost path based on the determined result, the target position of the UAV, and the improved artificial potential field method, including: If the determination result indicates that the UAV is not currently in the preset scenario, the lowest cost path location is searched based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method under the current flight parameters of the UAV. or, If the determination result indicates that the UAV is currently in a preset scenario, the current flight parameters of the UAV are corrected, and under the corrected flight parameters, the lowest cost path position is searched based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method.

[0078] For example, in an embodiment of this application, the search unit 403 is used to correct the current flight parameters of the drone when the determination result indicates that the drone is currently in a preset scenario, including: When the drone is currently in a scenario of altitude change, the drone's current flight altitude is corrected based on the width of the vertical region of the drone's current flight path. or, When the drone is currently in a scenario with a single wall, the corresponding straight line model of the wall is determined based on the RANSAC algorithm, and the current main heading of the drone is corrected based on the straight line model of the wall.

[0079] For example, in an embodiment of this application, the search unit 403 is used to determine the corresponding wall straight line model based on the RANSAC algorithm, including: Based on the current main heading of the UAV and the local grid map, obtain a set of wall edge points for navigation determination; For the set of wall edge points, the initial straight-line model wall with the most interior points is determined based on the RANSAC algorithm; The wall straight model is determined based on the vertical distance between each interior point and the initial straight model wall.

[0080] For example, in an embodiment of this application, the search unit 403 is used to determine the wall straight model based on the vertical distance between each interior point and the initial straight model wall, including: For each interior point, the weight corresponding to the interior point is determined based on the vertical distance between the interior point and the wall of the initial straight line model; Based on the vertical distance and weight of each interior point, a target loss function is constructed. The target loss function is used to characterize the degree of matching between all interior points and the initial straight line model wall. The target loss function is solved using the weighted least squares method until the number of iterations reaches a preset threshold, thus obtaining the straight line model of the wall.

[0081] For example, in an embodiment of this application, the search unit 403 is used to search for the lowest-cost path location based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method, including: Based on the current position and target position of the UAV, the repulsive and attractive forces corresponding to the current position are calculated using the improved artificial potential field method. Based on the repulsive and attractive forces, a target navigation area is determined from the local grid map, the target navigation area including multiple grids; For each grid, the passage cost value corresponding to the center point of the grid is determined, wherein the smaller the passage cost value, the farther the center point is from the obstacle; The center point of the grid corresponding to the minimum passage cost value among the passage cost values ​​of the center point of each grid is determined as the location of the lowest cost path.

[0082] The fully autonomous UAV navigation device 40 for underground utility tunnel environments provided in this application embodiment can execute the technical solution of the fully autonomous UAV navigation method for underground utility tunnel environments in any of the above embodiments. Its implementation principle and beneficial effects are similar to those of the fully autonomous UAV navigation method for underground utility tunnel environments. Please refer to the implementation principle and beneficial effects of the fully autonomous UAV navigation method for underground utility tunnel environments. It will not be repeated here.

[0083] Figure 5 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of this application, such as... Figure 5 As shown, the electronic device may include a processor 510, a communications interface 520, a memory 530, and a communication bus 540, wherein the processor 510, communications interface 520, and memory 530 communicate with each other via the communication bus 540. The processor 510 can call logical instructions in the memory 530 to execute a fully autonomous navigation method for an underground utility tunnel environment. This method includes: constructing a local grid map corresponding to the current location of the UAV in the underground utility tunnel environment based on the UAV's current position, the UAV's size, and the radius of curvature of the UAV's current flight trajectory; determining whether the UAV is currently in a preset scenario based on the UAV's current position and the local grid map, the preset scenario including a height change scenario or a single-sided wall scenario; searching for the lowest-cost path location based on the determination result, the UAV's current position, the UAV's target position, and an improved artificial potential field method; and performing fully autonomous navigation based on the lowest-cost path location if the lowest-cost path location is found.

[0084] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0085] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the fully autonomous navigation method for UAVs in underground utility tunnel environments provided by the above methods. The method includes: constructing a local grid map corresponding to the current location of the UAV in the underground utility tunnel environment based on the current location of the UAV, the size of the UAV, and the radius of curvature of the current flight trajectory of the UAV; determining whether the UAV is currently in a preset scenario based on the current location of the UAV and the local grid map, the preset scenario including a height change scenario or a single-sided wall scenario; searching for the lowest-cost path location based on the determination result, the current location of the UAV, the target location of the UAV, and an improved artificial potential field method; and performing fully autonomous navigation based on the lowest-cost path location if the lowest-cost path location is found.

[0086] Furthermore, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a fully autonomous navigation method for an underground utility tunnel environment provided by the methods described above. This method includes: constructing a local grid map corresponding to the current location of the UAV in the underground utility tunnel environment based on the UAV's current position, the UAV's size, and the radius of curvature of the UAV's current flight trajectory; determining whether the UAV is currently in a preset scenario based on the UAV's current position and the local grid map, the preset scenario including a height change scenario or a single-sided wall scenario; searching for a minimum-cost path location based on the determination result, the UAV's current position, the UAV's target position, and an improved artificial potential field method; and performing fully autonomous navigation based on the minimum-cost path location if the minimum-cost path location is found.

[0087] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments, characterized in that, include: Based on the drone's current location, the drone's size, and the radius of curvature of the drone's current flight trajectory, a local grid map corresponding to the drone's current location in the underground utility tunnel environment is constructed. Based on the current position of the drone and the local grid map, it is determined whether the drone is currently in a preset scene, which includes a scene with changes in altitude or a scene with a single wall. Based on the determined results, the current position of the UAV, the target position of the UAV, and the improved artificial potential field method, the lowest cost path position is searched. Once the location of the lowest-cost path is determined, fully autonomous navigation is performed based on that location.

2. The method according to claim 1, characterized in that, The search for the lowest-cost path based on the determined result, the target position of the UAV, and the improved artificial potential field method includes: If the determination result indicates that the UAV is not currently in the preset scenario, the lowest cost path location is searched based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method under the current flight parameters of the UAV. or, If the determination result indicates that the UAV is currently in a preset scenario, the current flight parameters of the UAV are corrected, and under the corrected flight parameters, the lowest cost path position is searched based on the current position of the UAV, the target position of the UAV, and the improved artificial potential field method.

3. The method according to claim 2, characterized in that, When the determination result indicates that the drone is currently in a preset scenario, the step of correcting the current flight parameters of the drone includes: When the drone is currently in a scenario of altitude change, the drone's current flight altitude is corrected based on the width of the vertical region of the drone's current flight path. or, When the drone is currently in a scenario with a single wall, the corresponding straight line model of the wall is determined based on the RANSAC algorithm, and the current main heading of the drone is corrected based on the straight line model of the wall.

4. The method according to claim 3, characterized in that, The determination of the corresponding straight wall model based on the RANSAC algorithm includes: Based on the current main heading of the UAV and the local grid map, obtain a set of wall edge points for navigation; For the set of wall edge points, the initial straight-line model wall with the most interior points is determined based on the RANSAC algorithm; The wall straight model is determined based on the vertical distance between each interior point and the initial straight model wall.

5. The method according to claim 4, characterized in that, The process of determining the straight wall model based on the vertical distance between each interior point and the initial straight wall model includes: For each interior point, the weight corresponding to the interior point is determined based on the vertical distance between the interior point and the wall of the initial straight line model; Based on the vertical distance and weight of each interior point, a target loss function is constructed. The target loss function is used to characterize the degree of matching between all interior points and the initial straight line model wall. The target loss function is solved using the weighted least squares method until the number of iterations reaches a preset threshold, thus obtaining the straight line model of the wall.

6. The method according to any one of claims 2-5, characterized in that, The search for the lowest-cost path location based on the current location of the UAV, the target location of the UAV, and the improved artificial potential field method includes: Based on the current position and target position of the UAV, the repulsive and attractive forces corresponding to the current position are calculated using the improved artificial potential field method. Based on the repulsive and attractive forces, a target navigation area is determined from the local grid map, the target navigation area including multiple grids; For each grid, the passage cost value corresponding to the center point of the grid is determined, wherein the smaller the passage cost value, the farther the center point is from the obstacle; The center point of the grid corresponding to the minimum passage cost value among the passage cost values ​​of the center point of each grid is determined as the location of the lowest cost path.

7. A fully autonomous navigation device for unmanned aerial vehicles (UAVs) in underground utility tunnel environments, characterized in that: include: The construction unit is used to construct a local grid map of the UAV in the current environment facing the underground utility tunnel, based on the UAV's current position, the UAV's size, and the radius of curvature of the UAV's current flight trajectory. The determining unit is used to determine whether the drone is currently in a preset scene based on the drone's current position and the local grid map. The preset scene includes a scene with changes in altitude or a scene with a single wall. The search unit is used to search for the lowest cost path location based on the determined results, the current position of the UAV, the target position of the UAV, and the improved artificial potential field method. The navigation unit is used to perform fully autonomous navigation based on the lowest-cost path location once the lowest-cost path location has been found.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the fully autonomous navigation method for unmanned aerial vehicles (UAVs) in underground utility tunnel environments as described in any one of claims 1 to 6.

Images