Dam slope-oriented unmanned vehicle navigation method, apparatus and device, and medium

Abstract

The invention discloses an unmanned vehicle navigation method and device facing a dam slope, equipment and a medium. Comprising the following steps: acquiring original position data, original attitude data and original three-dimensional point cloud data acquired by a multi-dimensional sensor; performing attitude compensation and slope geometric distortion correction processing on the original three-dimensional point cloud data to obtain a high-precision point cloud and dam slope three-dimensional grid model; carrying out simultaneous localization and map construction optimization by taking gradient information as a core constraint, and generating a high-precision global pose and a global reference navigation path; calculating a dynamic feasible region map in combination with preset vehicle dynamics constraints; planning to obtain a local obstacle avoidance path by taking the high-precision global pose as a planning starting point and the global reference navigation path as guidance; and converting the local obstacle avoidance path into a control instruction to drive the unmanned vehicle to automatically drive according to the control instruction. According to the invention, high-precision, safe and stable autonomous inspection of the unmanned vehicle in a dam slope environment is realized, and reliable technical support is provided for safe operation and maintenance of a dam.

Description

Technical Field

[0001] This invention relates to the field of automated operation and maintenance technology, and is applicable to the field of unmanned inspection of dikes in water conservancy projects. In particular, it relates to an unmanned vehicle navigation method, device, equipment and medium for dike slopes. Background Technology

[0002] As a core flood control and water supply infrastructure, the safe operation of dikes directly affects the livelihoods of people in the basin and the stability of the regional economy. Traditional manual dike inspections are inefficient, time-consuming, and extremely risky in complex terrain and inclement weather. Therefore, unmanned intelligent inspection has become an industry trend. Existing unmanned vehicle navigation systems are mostly designed based on the assumption of level ground, but dikes have special terrain such as slopes of 15° to 35°. Unmanned vehicles need to move continuously between flat and sloping areas, which existing systems cannot adapt to and cannot meet the requirements for accurate navigation.

[0003] Existing autonomous vehicle navigation systems are ill-suited for sloping embankment environments, exhibiting significant core shortcomings: Firstly, traditional navigation systems, based on the assumption of flat terrain, are prone to point cloud distortion, attitude calculation deviations, and cumulative mileage errors in sloping environments using Simultaneous Localization and Mapping (SLAM) algorithms. GPS is susceptible to multipath effects due to terrain obstruction, while IMUs are exacerbated by slope tilt and bumps, leading to increased positioning drift. Secondly, in transitional areas between flat and sloping terrain, the combined effects of these problems result in uncontrolled positioning accuracy, ultimately increasing the risk of autonomous vehicles deviating from their paths and becoming unstable, thus failing to support practical applications for unmanned embankment inspection. Summary of the Invention

[0004] Based on this, the present invention provides an unmanned vehicle navigation method, device, equipment and medium for dam slopes, to solve the problem that traditional navigation methods based on the flat terrain assumption are difficult to adapt to the special non-horizontal terrain of dam slopes and cannot support unmanned, accurate and safe inspection of dams.

[0005] In a first aspect, embodiments of the present invention provide a navigation method for unmanned vehicles facing a dam slope, comprising:

[0006] Acquire the raw position data and attitude data of the unmanned vehicle, as well as the raw three-dimensional point cloud data of the embankment slope environment to be inspected, collected by the unmanned vehicle using multi-dimensional sensors.

[0007] The original three-dimensional point cloud data is subjected to attitude compensation and slope geometric distortion correction to obtain a high-precision point cloud. Based on the high-precision point cloud, a three-dimensional mesh model of the dam slope representing the actual terrain of the dam slope is generated.

[0008] Using the slope information in the three-dimensional mesh model of the dam slope as the core constraint, and combining the original pose data and the original position data, real-time positioning and map building optimization are performed to generate a high-precision global pose and a global reference navigation path with slope constraints.

[0009] Based on the high-precision point cloud and the three-dimensional mesh model of the dam slope, combined with the preset vehicle dynamics constraints, a dynamic feasible domain map is calculated.

[0010] Using the high-precision global pose as the planning starting point and the global reference navigation path as the guide, a local obstacle avoidance path is planned within the dynamic feasible domain map.

[0011] The local obstacle avoidance path is converted into control commands and sent to the actuators of the unmanned vehicle in real time to drive the unmanned vehicle to drive automatically according to the control commands.

[0012] Secondly, embodiments of the present invention also provide an unmanned vehicle navigation device facing a dam slope, comprising:

[0013] The raw data acquisition module is used to acquire the raw position data of the unmanned vehicle, the raw attitude data of the unmanned vehicle, and the raw three-dimensional point cloud data of the slope environment of the embankment to be inspected, which are collected by the unmanned vehicle using multi-dimensional sensors.

[0014] The attitude compensation and geometric correction module is used to perform attitude compensation and slope geometric distortion correction on the original three-dimensional point cloud data to obtain a high-precision point cloud, and generate a three-dimensional mesh model of the dam slope representing the actual terrain of the dam slope based on the high-precision point cloud.

[0015] The real-time positioning and map building optimization module is used to perform real-time positioning and map building optimization by taking the slope information in the three-dimensional mesh model of the dam slope as the core constraint, and combining the original attitude data and the original position data to generate a high-precision global pose and a global reference navigation path with slope constraints.

[0016] The dynamic feasible domain map calculation module is used to calculate the dynamic feasible domain map based on the high-precision point cloud and the three-dimensional mesh model of the dam slope, combined with preset vehicle dynamics constraints.

[0017] The local obstacle avoidance path planning module is used to plan a local obstacle avoidance path within the dynamic feasible domain map, with the high-precision global pose as the planning starting point and the global reference navigation path as the guide.

[0018] The unmanned vehicle drive module is used to convert the local obstacle avoidance path into control commands and send them to the actuators of the unmanned vehicle in real time, so as to drive the unmanned vehicle to drive automatically according to the control commands.

[0019] Thirdly, embodiments of the present invention also provide an electronic device, the electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to execute an unmanned vehicle navigation method for a dam slope as described in any embodiment of the present invention.

[0020] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing computer instructions, which are used to cause a processor to execute and implement the unmanned vehicle navigation method for a dam slope as described in any embodiment of the present invention.

[0021] This invention, through attitude compensation and slope geometric distortion correction of the original point cloud, accurately restores the terrain features of the embankment slope, providing a reliable foundation for positioning and navigation, and effectively reducing positioning errors caused by terrain perception deviations. By combining slope information to construct core constraints and optimize real-time positioning and map building, it significantly suppresses attitude drift in slope environments, achieving high-precision global positioning for unmanned vehicles. Furthermore, based on terrain features and vehicle dynamics constraints, a dynamic feasible domain map is generated, and a local obstacle avoidance path is planned guided by a global reference path, ensuring stable driving and autonomous obstacle avoidance of the unmanned vehicle in complex slope environments. Ultimately, without human intervention, the unmanned vehicle can continuously complete embankment slope inspections, significantly improving inspection efficiency, coverage, and operational safety, breaking through the application limitations of traditional flat-ground navigation methods, and providing reliable technical support for embankment safety operation and maintenance.

[0022] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart of an unmanned vehicle navigation method for a dam slope according to Embodiment 1 of the present invention;

[0025] Figure 2 This is a flowchart of another unmanned vehicle navigation method for a dam slope provided in Embodiment 2 of the present invention;

[0026] Figure 3 This is a schematic diagram of the structure of an unmanned vehicle navigation device facing a dam slope according to Embodiment 3 of the present invention;

[0027] Figure 4 This is a schematic diagram of the structure of an electronic device that implements an unmanned vehicle navigation method for a dam slope according to an embodiment of the present invention. Detailed Implementation

[0028] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0030] Example 1

[0031] Figure 1 This is a flowchart of an unmanned vehicle navigation method for a dam slope according to Embodiment 1 of the present invention. This embodiment is applicable to situations where unmanned vehicles conduct autonomous safety inspections of dam slope areas. The method can be executed by an unmanned vehicle navigation device for a dam slope. This device can be implemented in hardware and / or software and can be configured in a navigation system compatible with the unmanned vehicle driving system. Figure 1 As shown, the method includes:

[0032] S110. Acquire the unmanned vehicle's original position data, original attitude data, and original three-dimensional point cloud data of the embankment slope environment to be inspected, collected by the unmanned vehicle using multi-dimensional sensors.

[0033] In this embodiment, the dam slope refers to the core component area of ​​a dam in a water conservancy project, encompassing the continuously connected dam crest plane, sloping dam slope, and transition area at the dam toe. It is a crucial component for the dam's flood control and water supply functions, and also the core operating area for unmanned vehicle (UAV) inspections. The UAV refers to an autonomous vehicle specifically designed for dam slope inspection tasks, integrating multi-dimensional sensing equipment such as 3D LiDAR, GNSS receiver, and attitude sensors, as well as a high-performance data processing platform. The dam slope environment to be inspected specifically refers to the specific dam section where the UAV plans to perform its inspection task, encompassing the dam crest, slope, dam toe, and surrounding environment of that section.

[0034] Multi-dimensional sensors operate synchronously according to a unified system clock signal, ensuring consistency in the timestamps of raw position data, raw attitude data, and raw 3D point cloud data. Due to the sloping terrain of the embankment, data asynchrony can amplify subsequent correction and positioning errors; synchronous acquisition is fundamental for adapting to the sloping environment. Raw position data provides global position anchors, raw attitude data reflects the vehicle's own state, and raw 3D point cloud data describes the external terrain environment, laying the data foundation for subsequent correction, positioning, obstacle avoidance, and other steps.

[0035] S120. Perform attitude compensation and slope geometric distortion correction on the original three-dimensional point cloud data to obtain a high-precision point cloud, and generate a three-dimensional mesh model of the dam slope representing the actual terrain of the dam slope based on the high-precision point cloud.

[0036] Attitude compensation refers to the process of transforming the original 3D point cloud data from the vehicle's body coordinate system to the world coordinate system based on the original attitude data, using coordinate transformation to eliminate the influence of the vehicle's own attitude changes on the original point cloud data. Slope geometric distortion correction refers to further optimizing the attitude-compensated point cloud data for the tilt characteristics of the embankment slope. The core of this process is to correct the geometric deviations of the point cloud generated by the LiDAR when acquiring data on the tilted surface through local slope normal vector optimization and gravity direction constraints. High-precision point cloud refers to 3D point cloud data that has undergone attitude compensation and slope geometric distortion correction, eliminating attitude interference and geometric deviations, and accurately reflecting the actual terrain features of the embankment slope. The 3D mesh model of the embankment slope refers to a continuous and smooth model representing the actual terrain of the embankment slope, generated based on the high-precision point cloud using a 3D surface reconstruction algorithm. It can intuitively present terrain information such as slope changes and concave-convex structures of the embankment slope.

[0037] In terms of processing order, the influence of the vehicle's own attitude fluctuations on the original point cloud is first eliminated through attitude compensation (for example, when the vehicle tilts on a slope, the point cloud will deviate due to the sensor attitude shift, which needs to be corrected by roll and pitch angle data); then, the systematic deviation of the point cloud caused by the slope inclination is specifically addressed through slope geometric distortion correction (for example, the angle between the slope normal vector and the gravity direction is optimized to ensure that the point cloud fits the actual slope).

[0038] S130. Using the slope information in the three-dimensional mesh model of the dam slope as the core constraint, and combining the original attitude data and the original position data, perform real-time positioning and map building optimization to generate a high-precision global pose and a global reference navigation path with slope constraints.

[0039] Slope information refers to the local slope data (i.e., the angle between the slope and the horizontal plane at that location) at each position in the 3D mesh model of the dam slope. Core constraints are the highest priority and most decisive constraints in the real-time localization and map building optimization process. In this embodiment, slope information is used as the core constraint to avoid attitude drift that occurs in traditional localization methods in sloping environments. Real-time localization and map building optimization refers to the process of dynamically optimizing the real-time position of the unmanned vehicle and the surrounding environment map by combining sensor data and constraints (i.e., SLAM optimization). The core is to improve positioning accuracy by minimizing errors. High-precision global pose with slope constraints refers to the accurate position and attitude of the unmanned vehicle in the global coordinate system obtained after incorporating slope constraints during the localization optimization process. The positioning accuracy is better than traditional methods without slope constraints. Global reference navigation path refers to the overall navigation route planned based on the terrain features of the 3D mesh model of the dam slope, covering the entire inspection area, used to guide local obstacle avoidance paths without deviating from the core inspection direction.

[0040] S140. Based on the high-precision point cloud and the three-dimensional mesh model of the dam slope, and combined with the preset vehicle dynamics constraints, calculate the dynamic feasible domain map.

[0041] Preset vehicle dynamics constraints refer to the pre-defined vehicle driving constraints based on the autonomous vehicle's own performance and the safety requirements for dam slope inspection. These constraints include the maximum permissible roll angle, slope friction coefficient, and minimum turning radius, and are used to ensure vehicle driving stability. The dynamic feasible domain map refers to a map of areas where the autonomous vehicle can safely drive, defined based on terrain information, obstacle information, and dynamic constraints. It is represented as a set of passable grid cells.

[0042] Based on high-precision point clouds and 3D mesh models of dam slopes, combined with preset vehicle dynamics constraints, the drivability of grid cells is determined by these three factors. Since the high-precision point cloud and vehicle status are updated in real time, the dynamic feasible domain map is updated synchronously. For example, when new obstacles or sudden changes in slope gradient are detected, the drivable area is adjusted in real time to ensure that the map is consistent with the actual environment, providing an accurate range reference for subsequent obstacle avoidance path planning.

[0043] S150. Using the high-precision global pose as the planning starting point and the global reference navigation path as the guide, a local obstacle avoidance path is planned within the dynamic feasible domain map.

[0044] Local obstacle avoidance path refers to a safe driving path planned for the current local environment within a dynamic feasible domain map to avoid obstacles. It is a local adjustment of the global reference navigation path. On the one hand, it starts with a high-precision global pose to ensure the accuracy of the initial position of the planned path; on the other hand, it is guided by the global reference navigation path to limit the extension direction of the local path, so that local obstacle avoidance does not deviate from the global path.

[0045] S160. The local obstacle avoidance path is converted into control commands and sent to the actuator of the unmanned vehicle in real time to drive the unmanned vehicle to drive automatically according to the control commands.

[0046] Control commands refer to the operational instructions that convert the local obstacle avoidance path into operable commands that the autonomous vehicle's actuators can recognize. These commands specifically include parameters such as steering angle, speed, and braking intensity. Actuators are the components in the autonomous vehicle responsible for performing driving operations, including the power system, steering system, and braking system, used to receive control commands and drive the vehicle's movement. Autonomous driving refers to the process by which an autonomous vehicle completes driving, obstacle avoidance, and path inspection autonomously based on control commands without human intervention, relying entirely on the decision-making output of the navigation system. Because the local obstacle avoidance path is dynamically updated, control commands are also sent synchronously in real time, ensuring that the autonomous vehicle can respond instantly to environmental changes (e.g., when encountering a sudden obstacle, the new obstacle avoidance path is quickly converted into control commands, driving the vehicle to adjust its direction), ultimately achieving autonomous and safe inspection in dam slope environments.

[0047] This invention, through attitude compensation and slope geometric distortion correction of the original point cloud, accurately restores the terrain features of the embankment slope, providing a reliable foundation for positioning and navigation, and effectively reducing positioning errors caused by terrain perception deviations. By combining slope information to construct core constraints and optimize real-time positioning and map building, it significantly suppresses attitude drift in slope environments, achieving high-precision global positioning for unmanned vehicles. Furthermore, based on terrain features and vehicle dynamics constraints, a dynamic feasible domain map is generated, and a local obstacle avoidance path is planned guided by a global reference path, ensuring stable driving and autonomous obstacle avoidance of the unmanned vehicle in complex slope environments. Ultimately, without human intervention, the unmanned vehicle can continuously complete embankment slope inspections, significantly improving inspection efficiency, coverage, and operational safety, breaking through the application limitations of traditional flat-ground navigation methods, and providing reliable technical support for embankment safety operation and maintenance.

[0048] Example 2

[0049] Figure 2 This is a flowchart of another unmanned vehicle navigation method for dam slopes provided in Embodiment 2 of the present invention. This embodiment is a refinement based on Embodiment 1, specifically as follows: Figure 2 As shown, the method includes:

[0050] S210. Acquire the unmanned vehicle's original position data, original attitude data, and original three-dimensional point cloud data of the embankment slope environment to be inspected, all collected by the unmanned vehicle using multi-dimensional sensors.

[0051] Optionally, acquiring the unmanned vehicle's raw position data, raw attitude data, and raw 3D point cloud data of the embankment slope environment to be inspected, collected by the unmanned vehicle using multi-dimensional sensors, may include:

[0052] Acquire the raw location data of the unmanned vehicle using its onboard GNSS receiver;

[0053] Acquire the raw attitude data of the autonomous vehicle collected by its onboard attitude sensors; and

[0054] Obtain raw 3D point cloud data of the embankment slope environment to be inspected, collected by an unmanned vehicle using an onboard 3D LiDAR.

[0055] A vehicle-mounted Global Navigation Satellite System (GNSS) receiver refers to a navigation and positioning device integrated into an unmanned vehicle. It is installed in an open location on the top of the unmanned vehicle, supports multi-frequency and multi-mode reception and RTK differential correction technology. Its core function is to receive satellite signals, collect the initial positioning data of the unmanned vehicle in the global coordinate system, and provide global position anchors for subsequent positioning optimization.

[0056] Vehicle attitude sensors are motion sensing devices integrated into autonomous vehicles. Installed in the vehicle chassis or suspension system, their core components include tilt sensors and gyroscopes. They collect data using high-frequency sampling and directly output the vehicle's roll, pitch, and yaw angles, reflecting real-time attitude changes when driving on slopes. The installation location can be chosen to be either the vehicle chassis or the suspension system because this position is closest to the core of the vehicle's motion, accurately capturing attitude changes such as roll and pitch when driving on embankment slopes (e.g., pitch angle fluctuations caused by sudden changes in slope gradient), avoiding attitude data distortion due to installation location deviations.

[0057] Vehicle-mounted 3D LiDAR refers to a 3D environmental perception device integrated into an unmanned vehicle. Mounted on a stable bracket above the vehicle body, it possesses 360° horizontal scanning capability and a vertical field of view. Its core function is to emit laser beams to scan the surrounding environment and generate 3D point cloud data by receiving reflected signals, comprehensively capturing the terrain undulations and obstacle distribution of the embankment slope to be inspected. The stable bracket mounted on the vehicle body, equipped with a vibration suppression module, is designed specifically for driving on embankment slopes. The stable bracket prevents sensor attitude shifts caused by vehicle tilting and bumps, while the vibration suppression module reduces interference from slope vibrations on laser emission and reception, ensuring the geometric accuracy of the original 3D point cloud data.

[0058] Furthermore, before acquiring the unmanned vehicle's raw position data, raw attitude data, and raw 3D point cloud data of the embankment slope environment to be inspected, which are collected by the unmanned vehicle using multi-dimensional sensors, the process may also include:

[0059] After detecting that the unmanned vehicle has completed power-on initialization, a data acquisition start command and a unified system clock signal are generated and sent to the unmanned vehicle;

[0060] The acquisition start command is used to instruct the unmanned vehicle to start collecting data using multi-dimensional sensors, and the unified system clock signal is used to instruct the unmanned vehicle to assign a unified timestamp to the collected raw position data, raw attitude data, and raw 3D point cloud data.

[0061] Power-on initialization refers to the process by which the navigation system and related hardware complete self-checks, parameter configurations, and status readiness after the unmanned vehicle is powered on. Specifically, this may include functional self-checks of multi-dimensional sensors, startup of the data processing platform, communication adaptation between sensors and the processing platform, and status calibration of actuators, ensuring that all equipment meets the data acquisition and navigation control requirements for dam slope inspection. The data acquisition start command is a trigger signal generated by the navigation system after power-on initialization. It is used to uniformly instruct the unmanned vehicle's multi-dimensional sensors to synchronously begin data acquisition, avoiding timing discrepancies caused by different sensors starting independently. A unified system clock signal is used to assign consistent time identifiers to various types of raw data acquired by the multi-dimensional sensors, ensuring time dimension alignment of data from different sources. A unified timestamp refers to a unique time identifier, based on the unified system clock signal, for each set of acquired raw position data, raw attitude data, and raw 3D point cloud data. This identifier is accurate to the instant of acquisition, ensuring that the three types of data at the same point in time can be accurately correlated and matched.

[0062] S220. Perform attitude compensation and slope geometric distortion correction on the original three-dimensional point cloud data to obtain a high-precision point cloud, and generate a three-dimensional mesh model of the dam slope representing the actual terrain of the dam slope based on the high-precision point cloud.

[0063] Optionally, the original 3D point cloud data is subjected to attitude compensation and slope geometric distortion correction processing to obtain a high-precision point cloud. Based on the high-precision point cloud, a 3D mesh model of the dam slope representing the actual terrain of the dam slope is generated, which may include:

[0064] Based on the original attitude data of the unmanned vehicle, coordinate transformation is performed on the original three-dimensional point cloud data to obtain the attitude-compensated point cloud.

[0065] The attitude compensation point cloud is divided into local neighborhoods to obtain multiple sub-neighborhoods, and the local slope normal vector is estimated for each sub-neighborhood.

[0066] Based on the gravity direction vector extracted from the original attitude data, an objective function for optimizing the angle between the gravity direction and the local slope normal vector is constructed, and the local slope normal vector is iteratively optimized based on the objective function for optimizing the angle.

[0067] Based on the iteratively optimized local slope normal vector, the attitude compensation point cloud is finely adjusted along the local slope normal direction to obtain a geometrically corrected high-precision point cloud, and a three-dimensional mesh model of the dam slope is generated based on the high-precision point cloud.

[0068] The core basis of coordinate transformation is the original attitude data. When an autonomous vehicle travels on a dam slope, the inclination of the slope causes the vehicle to continuously roll and pitch. However, since the LiDAR is fixed to the vehicle body, the original point cloud will be offset and does not reflect the actual terrain features. By transforming the point cloud from the vehicle coordinate system to the world coordinate system using a rotation matrix, the offset caused by the vehicle's attitude can be corrected in reverse, allowing the point cloud to initially return to its true position on the terrain.

[0069] Local neighborhood partitioning refers to dividing the attitude-compensated point cloud into several non-overlapping small-scale point cloud sets (i.e., sub-neighborhoods) according to a preset spatial range. Each sub-neighborhood corresponds to a local terrain region on the embankment slope, ensuring independent analysis of slope characteristics at different locations. A sub-neighborhood is a single small-scale point cloud set obtained after local neighborhood partitioning. Its spatial range is small enough that the terrain within each sub-neighborhood can be approximated as a plane with a single slope, providing a premise for normal vector estimation. The local slope normal vector is a vector estimated through mathematical algorithms, perpendicular to the local slope corresponding to the sub-neighborhood. Its direction directly reflects the tilt and slope magnitude of the local slope and is a core parameter describing the slope's geometric characteristics. The slope of a dam is not uniform globally. Different sections of the dam, and different locations within the same section, may have abrupt changes in slope or gentle transitions. If the global normal vector is estimated for the entire point cloud, the local slope differences will be averaged, causing the normal vector to fail to reflect the true terrain. By dividing the terrain into local neighborhoods, the complex terrain can be decomposed into multiple local approximate planes, ensuring that the normal vector of each sub-neighborhood can accurately match the actual slope of its corresponding location.

[0070] For each sub-neighborhood, algorithms such as least squares plane fitting are used to obtain the normal vector of the plane by fitting the optimal plane of all point clouds in the sub-neighborhood. The normal vector directly represents the tilt direction and tilt degree of the local slope, providing a quantitative slope geometry basis for subsequent correction, especially suitable for the terrain features of dam slopes that are "continuous overall but varied locally".

[0071] The gravity direction vector is a vertically downward reference vector extracted from the original attitude data. Its direction is fixed and it serves as the absolute reference standard for determining whether the slope normal vector conforms to the actual terrain. The angle optimization objective function is a mathematical function constructed with the goal of ensuring that the angle between the gravity direction vector and the local slope normal vector conforms to the actual terrain slope. Its core is to iteratively adjust the direction of the normal vectors to make the angle between them consistent with the actual slope angle corresponding to the sub-neighborhood, thus eliminating errors in normal vector estimation. Iterative optimization refers to the process of repeatedly substituting data to calculate the angle optimization objective function and adjusting the direction of the local slope normal vector until the function value reaches its minimum, ensuring the accuracy of normal vector estimation.

[0072] Fine-tuning along the local slope normal direction refers to making minor positional adjustments to each point in the attitude-compensated point cloud along the normal direction of its sub-neighborhood, based on the iteratively optimized local slope normal vector. The core purpose is to correct the systematic geometric distortion of the point cloud caused by slope inclination. High-precision point cloud after geometric correction refers to point cloud data that, after attitude compensation and normal vector optimization fine-tuning, completely eliminates vehicle attitude interference and systematic distortion caused by slope inclination, and accurately restores the terrain's inherent convex and concave features. 3D mesh model generation of the dam slope refers to the process of transforming discrete point cloud data into a continuous, smooth 3D mesh structure based on high-precision point cloud data using a 3D surface reconstruction algorithm, intuitively presenting the overall terrain morphology and local details of the dam slope.

[0073] S230. Based on the original location data, obtain the corresponding theoretical slope information from the three-dimensional mesh model of the dam slope, and define the theoretical slope information as a slope consistency constraint condition.

[0074] Theoretical slope information refers to the preset slope data corresponding to the real-time position of the unmanned vehicle (RV) as represented by the original location data, retrieved from the 3D mesh model of the embankment slope. It serves as a reference standard for judging whether the actual driving slope of the RV conforms to the terrain. The slope consistency constraint takes "the actual driving slope of the RV maintaining consistency with the theoretical slope information" as the core constraint criterion. This constraint limits the direction of real-time localization and map building optimization, preventing the localization results from deviating from the actual terrain of the embankment slope. Since the 3D mesh model of the embankment slope is a model generated based on high-precision point clouds that truly reflects the terrain, each location corresponds to unique and accurate slope data. Combined with the original location data, the slope of the corresponding area in the model can be directly locked. This slope is the theoretical slope, an inherent attribute of the terrain itself, unaffected by sensor errors or vehicle motion.

[0075] S240. Construct observation residual terms based on the original 3D point cloud data, construct motion prediction residual terms based on the original attitude data, and construct slope consistency constraint residual terms based on the slope consistency constraint conditions.

[0076] The observation residual term, constructed based on the original 3D point cloud data, reflects the deviation between the lidar observations and the local map predictions. A smaller deviation indicates higher geometric accuracy in local positioning. The motion prediction residual term, constructed based on the original attitude data, reflects the deviation between the actual motion state and the predicted motion state of the autonomous vehicle. A smaller deviation indicates a closer consistency between the positioning result and the vehicle's own motion logic. The slope consistency constraint residual term, constructed based on slope consistency constraints, reflects the deviation between the actual driving slope of the autonomous vehicle and the theoretical slope information. A smaller deviation indicates a closer fit between the positioning result and the terrain slope characteristics.

[0077] The observation residual term focuses on the accuracy of environmental perception, adapting to the complex terrain features of embankment slopes; the motion prediction residual term focuses on the consistency of the vehicle's own state, adapting to the scenario of frequent changes in vehicle posture when driving on embankment slopes; the slope consistency constraint residual term focuses on terrain slope adaptability, directly addressing the core pain point of traditional positioning in slope environments.

[0078] S250. Based on the observation residual, motion prediction residual and slope consistency constraint residual, construct a weighted summation objective function, and calculate the local optimized pose by minimizing the objective function.

[0079] The weighted summation objective function is a mathematical function obtained by superimposing the observation residuals, motion prediction residuals, and slope consistency constraint residuals according to preset weights. The weights can be adjusted according to the complexity of the embankment slope environment. The core is to comprehensively balance the three types of errors and minimize the overall positioning error. Minimizing the objective function means solving for the minimum value of the objective function through a mathematical optimization algorithm. At this point, the comprehensive deviation of the three types of residuals is minimized, corresponding to the optimal overall accuracy of the positioning result. Locally optimized pose refers to the precise pose obtained by minimizing the objective function in the local area where the autonomous vehicle is currently traveling.

[0080] To address the environmental characteristics of the dam slope, weights can be flexibly adjusted. For example, in areas with abrupt slope changes, the weight of the slope consistency constraint residual term can be increased to ensure that the positioning prioritizes conforming to the terrain slope. In areas with dense obstacles, the weight of the observation residual term can be increased to ensure consistency between the positioning and environmental perception results. This design enables the positioning optimization to have environmental adaptability, adapting to the terrain differences in different areas of the dam slope.

[0081] S260. The local optimized pose is fused with the original position data to obtain a high-precision global pose. Based on the terrain continuity between the high-precision global pose and the 3D mesh model of the dam slope, a global reference navigation path that fits the terrain of the dam slope is planned and generated.

[0082] High-precision global pose refers to the pose obtained by integrating locally optimized pose with the original position data through a data fusion algorithm. It combines local positioning accuracy with global position consistency, and can accurately represent the actual position and attitude of the unmanned vehicle in the global coordinate system. Terrain continuity refers to the terrain features presented by the 3D mesh model of the dam slope, including the smooth transition and gradual slope changes between the dam top, dam slope, and dam toe. It is the core basis for planning the global reference navigation path. The global reference navigation path that conforms to the dam slope terrain refers to the navigation route covering the entire inspection area, planned based on high-precision global pose and terrain continuity. Its characteristics are that the route follows the terrain undulations and the slope changes gently, avoiding paths that conflict with the terrain.

[0083] While locally optimized pose offers high local accuracy, it may lead to global shifts over time; while raw position data provides good global consistency, its accuracy is susceptible to occlusion. A fusion algorithm can complement the advantages of both. Specifically, in areas with good GNSS signal, the raw position data is used as the core for calibrating the locally optimized pose; in areas with signal obstruction, the locally optimized pose is used to maintain positioning accuracy. This ensures that the unmanned vehicle maintains accurate global positioning during long-distance embankment slope inspections, adapting to long, cross-regional embankment inspection scenarios.

[0084] S270. Based on the high-precision point cloud and the three-dimensional mesh model of the dam slope, and combined with the preset vehicle dynamics constraints, calculate the dynamic feasible domain map.

[0085] Furthermore, based on the high-precision point cloud and the 3D mesh model of the dam slope, combined with preset vehicle dynamics constraints, a dynamic feasible domain map is calculated, which may include:

[0086] The high-precision point cloud is rasterized to obtain multiple grid cells. By calculating the distance between each grid cell and the current real-time position of the unmanned vehicle represented by the high-precision global pose, the obstacle distribution information in each grid cell is identified.

[0087] The local slope information and slope normal vector corresponding to each grid cell are queried from the three-dimensional mesh model of the dam slope to obtain the local terrain slope characteristics of each grid cell.

[0088] Based on the obstacle distribution information and local terrain slope characteristics, combined with preset vehicle dynamics constraints, the passability of each grid cell is determined independently, and a dynamic feasible domain map is generated using all grid cells determined to be passable.

[0089] A grid cell refers to an independent small area unit obtained by uniformly dividing a dam slope area covered by a high-precision point cloud according to a preset fixed spatial size. Each cell corresponds to a fixed range on the terrain, facilitating independent determination of the accessibility of a local area. Obstacle distribution information refers to a dataset formed by integrating information such as "whether there are obstacles" and "the relative distance between obstacles and the autonomous vehicle" from all grid cells. Local terrain slope features refer to the comprehensive features of the local slope information and slope normal vector corresponding to each grid cell. These are core parameters reflecting the terrain tilt and orientation of the grid cell, directly determining the driving stability of the autonomous vehicle in that area.

[0090] The preset vehicle dynamics constraints refer to the quantitative constraints pre-set based on the driving performance of the unmanned vehicle and the safety requirements of embankment slope inspection. These constraints are used to determine whether the unmanned vehicle faces risks such as rollover or skidding when driving within a specific grid cell. The drivability determination refers to the process of independently judging whether each grid cell meets the safe driving conditions for the unmanned vehicle, combining obstacle distribution information, local terrain slope characteristics, and vehicle dynamics constraints. The determination result is simply either drivable or impassable. The dynamic feasible domain map is a map formed by combining all the grid cells determined to be drivable according to their spatial location. It visually presents the area where the unmanned vehicle can safely drive in the current environment. Its dynamic nature is reflected in real-time adjustments as high-precision point clouds and vehicle positions are updated. Each grid cell is determined independently, ensuring that the impassability of a certain area does not affect the overall map generation. The dynamic feasible domain map is updated in real time with environmental changes, which can cope with sudden obstacles or slope changes encountered during embankment slope inspection, ensuring that the feasible domain range is always consistent with the actual environment and providing a reliable spatial reference for subsequent path planning.

[0091] S280. Using the high-precision global pose as the planning starting point and the global reference navigation path as the guide, a local obstacle avoidance path is planned within the dynamic feasible domain map.

[0092] Optionally, using the high-precision global pose as the planning starting point and the global reference navigation path as the guide, a local obstacle avoidance path can be planned within the dynamic feasible domain map, which may include:

[0093] Using the real-time position represented by the high-precision global pose as the starting position for path planning, and using the global reference navigation path as a constraint, a comprehensive cost function is constructed.

[0094] Within the dynamic feasible region map, a local obstacle avoidance path is planned by minimizing the comprehensive cost function.

[0095] The comprehensive cost function is a mathematical function that integrates multiple cost indicators such as path length, slope adaptability, and obstacle avoidance safety. Its core function is to quantify the merits of different candidate paths; a smaller value indicates a path that better meets the inspection requirements. The planned starting position refers to the real-time position of the unmanned vehicle, accurately represented by its high-precision global pose. This serves as the starting point for path planning, ensuring the path originates from the vehicle's actual location without any initial deviation. Using the global reference navigation path as a constraint means applying the extension direction of the global reference navigation path and terrain adaptation principles as limiting conditions for path planning. This ensures that local obstacle avoidance paths do not deviate from the overall inspection route and always adhere to the main inspection line of the embankment slope.

[0096] Finding the minimum value of a function through mathematical optimization methods such as graph optimization essentially involves finding a balanced path that simultaneously satisfies the requirements of "shortest path, most stable slope, and safest obstacle avoidance" within the safe boundary of a dynamic feasible region map. For example, in steep slope areas, the algorithm prioritizes minimizing the slope deviation cost to ensure a gentle slope change; in areas with dense obstacles, it prioritizes minimizing the obstacle avoidance distance cost to ensure a safe distance from obstacles, accurately adapting to the environmental requirements of different areas of the dam slope.

[0097] S290. The local obstacle avoidance path is converted into control commands and sent to the actuator of the unmanned vehicle in real time to drive the unmanned vehicle to drive automatically according to the control commands.

[0098] This embodiment focuses on a detailed explanation of the specific implementation methods for data acquisition, point cloud correction, positioning optimization, dynamic feasible region calculation, and local obstacle avoidance path planning. It ensures the synchronization of multi-source data acquisition, the targeted nature of point cloud correction, the effectiveness of positioning optimization constraints, the comprehensiveness of feasible region determination, and the adaptability of path planning, making high-precision navigation and safety inspection in dam slope environments more practical for engineering applications.

[0099] Example 3

[0100] Figure 3 This is a schematic diagram of the structure of an unmanned vehicle navigation device facing a dam slope, provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes:

[0101] The raw data acquisition module 310 is used to acquire the raw position data of the unmanned vehicle, the raw attitude data of the unmanned vehicle, and the raw three-dimensional point cloud data of the slope environment of the embankment to be inspected, which are collected by the unmanned vehicle using multi-dimensional sensors.

[0102] The attitude compensation and geometric correction module 320 is used to perform attitude compensation and slope geometric distortion correction on the original three-dimensional point cloud data to obtain a high-precision point cloud, and generate a three-dimensional mesh model of the dam slope representing the actual terrain of the dam slope based on the high-precision point cloud.

[0103] The real-time positioning and map building optimization module 330 is used to perform real-time positioning and map building optimization by taking the slope information in the three-dimensional mesh model of the dam slope as the core constraint, and combining the original attitude data and the original position data to generate a high-precision global pose and a global reference navigation path with slope constraints.

[0104] The dynamic feasible domain map calculation module 340 is used to calculate the dynamic feasible domain map based on the high-precision point cloud and the three-dimensional mesh model of the dam slope, combined with preset vehicle dynamics constraints.

[0105] The local obstacle avoidance path planning module 350 is used to plan a local obstacle avoidance path within the dynamic feasible domain map, with the high-precision global pose as the planning starting point and the global reference navigation path as the guide.

[0106] The unmanned vehicle drive module 360 ​​is used to convert the local obstacle avoidance path into control commands and send them to the actuators of the unmanned vehicle in real time, so as to drive the unmanned vehicle to drive automatically according to the control commands.

[0107] This invention, through attitude compensation and slope geometric distortion correction of the original point cloud, accurately restores the terrain features of the embankment slope, providing a reliable foundation for positioning and navigation, and effectively reducing positioning errors caused by terrain perception deviations. By combining slope information to construct core constraints and optimize real-time positioning and map building, it significantly suppresses attitude drift in slope environments, achieving high-precision global positioning for unmanned vehicles. Furthermore, based on terrain features and vehicle dynamics constraints, a dynamic feasible domain map is generated, and a local obstacle avoidance path is planned guided by a global reference path, ensuring stable driving and autonomous obstacle avoidance of the unmanned vehicle in complex slope environments. Ultimately, without human intervention, the unmanned vehicle can continuously complete embankment slope inspections, significantly improving inspection efficiency, coverage, and operational safety, breaking through the application limitations of traditional flat-ground navigation methods, and providing reliable technical support for embankment safety operation and maintenance.

[0108] Optionally, based on the above embodiments, the raw data acquisition module 310 may include:

[0109] The location data acquisition unit is used to acquire the raw location data of the unmanned vehicle collected by the onboard GNSS receiver of the vehicle-mounted global navigation satellite system;

[0110] An attitude data acquisition unit is used to acquire the raw attitude data of the autonomous vehicle collected by its onboard attitude sensors; and,

[0111] The point cloud data acquisition unit is used to acquire the original three-dimensional point cloud data of the embankment slope environment to be inspected, which is collected by the unmanned vehicle using the vehicle-mounted three-dimensional lidar.

[0112] Optionally, based on the above embodiments, it may also include: a timestamp unification unit, used to generate a data acquisition start command and a unified system clock signal and send them to the unmanned vehicle before acquiring the unmanned vehicle's original position data, original attitude data and original three-dimensional point cloud data of the slope environment of the embankment to be inspected collected by the unmanned vehicle using multi-dimensional sensors; after detecting that the unmanned vehicle has completed power-on initialization.

[0113] The acquisition start command is used to instruct the unmanned vehicle to start collecting data using multi-dimensional sensors, and the unified system clock signal is used to instruct the unmanned vehicle to assign a unified timestamp to the collected raw position data, raw attitude data, and raw 3D point cloud data.

[0114] Optionally, based on the above embodiments, the attitude compensation and geometry correction module 320 may include:

[0115] The attitude compensation unit is used to perform coordinate transformation on the original three-dimensional point cloud data based on the original attitude data of the unmanned vehicle to obtain the attitude-compensated point cloud.

[0116] The domain partitioning unit is used to partition the attitude compensation point cloud into local neighborhoods, obtain multiple sub-neighborhoods, and estimate the local slope normal vector for each sub-neighborhood.

[0117] The objective function optimization unit is used to construct an objective function for optimizing the angle between the gravity direction and the local slope normal vector based on the gravity direction vector extracted from the original attitude data, and to iteratively optimize the local slope normal vector based on the objective function for optimizing the angle.

[0118] The geometric correction unit is used to fine-tune the attitude compensation point cloud along the local slope normal direction based on the iteratively optimized local slope normal vector to obtain a geometrically corrected high-precision point cloud, and generate a three-dimensional mesh model of the dam slope based on the high-precision point cloud.

[0119] Optionally, based on the above embodiments, the real-time positioning and map building optimization module 330 may include:

[0120] The slope consistency constraint definition unit is used to obtain the corresponding theoretical slope information from the three-dimensional mesh model of the dam slope based on the original location data, and define the theoretical slope information as slope consistency constraint conditions.

[0121] The residual term construction unit is used to construct observation residual terms based on the original 3D point cloud data, construct motion prediction residual terms based on the original attitude data, and construct slope consistency constraint residual terms based on the slope consistency constraint conditions.

[0122] The local optimization pose calculation unit is used to construct a weighted summation objective function based on the observation residual, motion prediction residual and slope consistency constraint residual, and calculate the local optimization pose by minimizing the objective function;

[0123] The global reference navigation path planning unit is used to fuse the local optimized pose with the original position data to obtain a high-precision global pose, and based on the terrain continuity between the high-precision global pose and the 3D mesh model of the dam slope, to plan and generate a global reference navigation path that fits the terrain of the dam slope.

[0124] Optionally, based on the above embodiments, the dynamic feasible domain map calculation module 340 may include:

[0125] The rasterization processing unit is used to perform rasterization processing on the high-precision point cloud to obtain multiple raster units, and to identify the obstacle distribution information in each raster unit by calculating the distance between each raster unit and the current real-time position of the unmanned vehicle represented by the high-precision global pose.

[0126] The local terrain slope feature acquisition unit is used to query the local slope information and slope normal vector corresponding to each grid cell from the three-dimensional mesh model of the dam slope, and obtain the local terrain slope feature of each grid cell.

[0127] The passability determination unit is used to independently determine the passability of each grid cell based on the obstacle distribution information and local terrain slope characteristics, combined with preset vehicle dynamics constraints, and to generate a dynamic feasible domain map using all grid cells determined to be passable.

[0128] Optionally, based on the above embodiments, the local obstacle avoidance path planning module 350 may include:

[0129] The comprehensive cost function construction unit is used to construct a comprehensive cost function with the real-time position represented by the high-precision global pose as the starting position for path planning and with the global reference navigation path as a constraint.

[0130] The comprehensive cost function minimization unit is used to plan a local obstacle avoidance path within the dynamic feasible domain map by minimizing the comprehensive cost function.

[0131] The unmanned vehicle navigation device facing the embankment slope provided in the embodiments of the present invention can execute the unmanned vehicle navigation method facing the embankment slope provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0132] Example 4

[0133] Figure 4A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0134] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0135] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0136] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as an unmanned vehicle navigation method for a dam slope.

[0137] That is: to acquire the original position data of the unmanned vehicle, the original attitude data of the unmanned vehicle, and the original three-dimensional point cloud data of the slope environment of the embankment to be inspected, all collected by the unmanned vehicle using multi-dimensional sensors;

[0138] The original three-dimensional point cloud data is subjected to attitude compensation and slope geometric distortion correction to obtain a high-precision point cloud. Based on the high-precision point cloud, a three-dimensional mesh model of the dam slope representing the actual terrain of the dam slope is generated.

[0139] Using the slope information in the three-dimensional mesh model of the dam slope as the core constraint, and combining the original pose data and the original position data, real-time positioning and map building optimization are performed to generate a high-precision global pose and a global reference navigation path with slope constraints.

[0140] Based on the high-precision point cloud and the three-dimensional mesh model of the dam slope, combined with the preset vehicle dynamics constraints, a dynamic feasible domain map is calculated.

[0141] Using the high-precision global pose as the planning starting point and the global reference navigation path as the guide, a local obstacle avoidance path is planned within the dynamic feasible domain map.

[0142] The local obstacle avoidance path is converted into control commands and sent to the actuators of the unmanned vehicle in real time to drive the unmanned vehicle to drive automatically according to the control commands.

[0143] In some embodiments, an unmanned vehicle navigation method for a dam slope can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the unmanned vehicle navigation method for a dam slope described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform an unmanned vehicle navigation method for a dam slope by any other suitable means (e.g., by means of firmware).

[0144] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0145] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0146] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0147] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0148] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0149] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0150] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0151] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A navigation method for unmanned vehicles facing a dam slope, characterized in that, The method includes: Acquire the raw position data and attitude data of the unmanned vehicle, as well as the raw three-dimensional point cloud data of the embankment slope environment to be inspected, collected by the unmanned vehicle using multi-dimensional sensors. The original three-dimensional point cloud data is subjected to attitude compensation and slope geometric distortion correction to obtain a high-precision point cloud. Based on the high-precision point cloud, a three-dimensional mesh model of the dam slope representing the actual terrain of the dam slope is generated. Using the slope information in the three-dimensional mesh model of the dam slope as the core constraint, and combining the original pose data and the original position data, real-time positioning and map building optimization are performed to generate a high-precision global pose and a global reference navigation path with slope constraints. Based on the high-precision point cloud and the three-dimensional mesh model of the dam slope, combined with the preset vehicle dynamics constraints, a dynamic feasible domain map is calculated. Using the high-precision global pose as the planning starting point and the global reference navigation path as the guide, a local obstacle avoidance path is planned within the dynamic feasible domain map. The local obstacle avoidance path is converted into control commands and sent to the actuators of the unmanned vehicle in real time to drive the unmanned vehicle to drive automatically according to the control commands.

2. The method according to claim 1, characterized in that, Acquire the raw position data and attitude data of the unmanned vehicle (UAV) collected by multi-dimensional sensors, as well as the raw 3D point cloud data of the embankment slope environment to be inspected, including: Acquire the raw location data of the unmanned vehicle using its onboard GNSS receiver; Acquire the raw attitude data of the autonomous vehicle collected by its onboard attitude sensors; and, Obtain raw 3D point cloud data of the embankment slope environment to be inspected, collected by an unmanned vehicle using an onboard 3D LiDAR.

3. The method according to claim 1, characterized in that, Before acquiring the unmanned vehicle's raw position data, raw attitude data, and raw 3D point cloud data of the embankment slope environment to be inspected, which are collected by the unmanned vehicle using multi-dimensional sensors, the following steps are also included: After detecting that the unmanned vehicle has completed power-on initialization, a data acquisition start command and a unified system clock signal are generated and sent to the unmanned vehicle; The acquisition start command is used to instruct the unmanned vehicle to start collecting data using multi-dimensional sensors, and the unified system clock signal is used to instruct the unmanned vehicle to assign a unified timestamp to the collected raw position data, raw attitude data, and raw 3D point cloud data.

4. The method according to any one of claims 1-3, characterized in that, The original 3D point cloud data is subjected to attitude compensation and slope geometric distortion correction to obtain a high-precision point cloud. Based on the high-precision point cloud, a 3D mesh model of the dam slope representing the actual terrain is generated, including: Based on the original attitude data of the unmanned vehicle, coordinate transformation is performed on the original three-dimensional point cloud data to obtain the attitude-compensated point cloud. The attitude compensation point cloud is divided into local neighborhoods to obtain multiple sub-neighborhoods, and the local slope normal vector is estimated for each sub-neighborhood. Based on the gravity direction vector extracted from the original attitude data, an objective function for optimizing the angle between the gravity direction and the local slope normal vector is constructed, and the local slope normal vector is iteratively optimized based on the objective function for optimizing the angle. Based on the iteratively optimized local slope normal vector, the attitude compensation point cloud is finely adjusted along the local slope normal direction to obtain a geometrically corrected high-precision point cloud, and a three-dimensional mesh model of the dam slope is generated based on the high-precision point cloud.

5. The method according to claim 1, characterized in that, Using the slope information in the 3D mesh model of the dam slope as the core constraint, and combining the original pose data and the original position data, real-time localization and map building optimization are performed to generate a high-precision global pose and a global reference navigation path with slope constraints, including: Based on the original location data, the corresponding theoretical slope information is obtained from the three-dimensional mesh model of the dam slope, and the theoretical slope information is defined as a slope consistency constraint condition. An observation residual term is constructed based on the original 3D point cloud data, a motion prediction residual term is constructed based on the original attitude data, and a slope consistency constraint residual term is constructed based on the slope consistency constraint condition. Based on the observation residual, motion prediction residual and slope consistency constraint residual, a weighted summation objective function is constructed, and the local optimal pose is calculated by minimizing the objective function. The local optimized pose is fused with the original position data to obtain a high-precision global pose. Based on the terrain continuity between the high-precision global pose and the 3D mesh model of the dam slope, a global reference navigation path that fits the terrain of the dam slope is planned and generated.

6. The method according to any one of claims 1-3, characterized in that, Based on the high-precision point cloud and the 3D mesh model of the dam slope, and combined with preset vehicle dynamics constraints, a dynamic feasible region map is calculated, including: The high-precision point cloud is rasterized to obtain multiple grid cells. By calculating the distance between each grid cell and the current real-time position of the unmanned vehicle represented by the high-precision global pose, the obstacle distribution information in each grid cell is identified. The local slope information and slope normal vector corresponding to each grid cell are queried from the three-dimensional mesh model of the dam slope to obtain the local terrain slope characteristics of each grid cell. Based on the obstacle distribution information and local terrain slope characteristics, combined with preset vehicle dynamics constraints, the passability of each grid cell is determined independently, and a dynamic feasible domain map is generated using all grid cells determined to be passable.

7. The method according to claim 6, characterized in that, Using the high-precision global pose as the planning starting point and the global reference navigation path as the guide, a local obstacle avoidance path is planned within the dynamic feasible domain map, including: Using the real-time position represented by the high-precision global pose as the starting position for path planning, and using the global reference navigation path as a constraint, a comprehensive cost function is constructed. Within the dynamic feasible domain map, a local obstacle avoidance path is planned by minimizing the comprehensive cost function.

8. A navigation device for unmanned vehicles facing a dam slope, characterized in that, The device includes: The raw data acquisition module is used to acquire the raw position data of the unmanned vehicle, the raw attitude data of the unmanned vehicle, and the raw three-dimensional point cloud data of the slope environment of the embankment to be inspected, which are collected by the unmanned vehicle using multi-dimensional sensors. The attitude compensation and geometric correction module is used to perform attitude compensation and slope geometric distortion correction on the original three-dimensional point cloud data to obtain a high-precision point cloud, and generate a three-dimensional mesh model of the dam slope representing the actual terrain of the dam slope based on the high-precision point cloud. The real-time positioning and map building optimization module is used to perform real-time positioning and map building optimization by taking the slope information in the three-dimensional mesh model of the dam slope as the core constraint, and combining the original attitude data and the original position data to generate a high-precision global pose and a global reference navigation path with slope constraints. The dynamic feasible domain map calculation module is used to calculate the dynamic feasible domain map based on the high-precision point cloud and the three-dimensional mesh model of the dam slope, combined with preset vehicle dynamics constraints. The local obstacle avoidance path planning module is used to plan a local obstacle avoidance path within the dynamic feasible domain map, with the high-precision global pose as the planning starting point and the global reference navigation path as the guide. The unmanned vehicle drive module is used to convert the local obstacle avoidance path into control commands and send them to the actuators of the unmanned vehicle in real time, so as to drive the unmanned vehicle to drive automatically according to the control commands.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform an unmanned vehicle navigation method for a dam slope according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute and implement the unmanned vehicle navigation method for a dam slope as described in any one of claims 1-7.

Images