Energy consumption optimization path planning method and system for unmanned aerial vehicle inspection

Abstract

The invention discloses an energy consumption optimization path planning method and system for unmanned aerial vehicle routing inspection, and relates to the technical field of unmanned aerial vehicle routing inspection path planning, and the method comprises the steps: calculating an evaluation function during node expansion in each step: a heuristic function represents the accumulated cost from a starting point to a current node, and a heuristic function represents the estimated cost from the node to a target point; the heuristic function is composed of a distance component and an energy consumption component, and the expression of the heuristic function is as follows: the linear distance from the current node to the target node, the minimum additional energy consumption estimated according to the height change and the flight resistance, and the weight coefficient; the improved heuristic function can preferentially avoid a high-energy-consumption area, a flatter and smoother air route is selected, and accumulated energy consumption and flight time of a global path are remarkably reduced; meanwhile, due to the introduction of the multi-dimensional cost, the global planning is more accurate and reliable, and the search efficiency of the algorithm in a complex three-dimensional environment is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle inspection path planning, and in particular to an energy consumption optimized path planning method and system for unmanned aerial vehicle inspection. Background Art

[0002] With the development of drone technology, autonomous drone inspections have gained widespread application in fields such as power and oil and gas pipeline inspections. Efficient and secure path planning is a key technology. Currently, commonly used path planning algorithms are categorized into two main categories: global planning and local obstacle avoidance, each with its own advantages and disadvantages. For global path planning, the classic A algorithm, due to its simplicity and efficiency, is widely used for shortest path optimization in two-dimensional environments. Some research has extended this algorithm to three-dimensional space for planning drone routes.

[0003] The Chinese patent with the announcement number CN119440049A discloses a method for generating a UAV inspection route based on the optimized A algorithm. This method improves the heuristic function of the A algorithm and f(n)=g(n)+α[h(n)+h(p)] This approach considers both the distance from the current node and the parent node to the target, avoiding areas with dense obstacles by integrating distance and obstacle density, and smoothing the path using the Floyd algorithm. This solution improves path safety and smoothness in static environments. However, while its heuristic function considers distance and obstacle distribution, it does not directly optimize actual drone energy consumption factors (such as climbing altitude and wind resistance).

[0004] Therefore, in high-speed flying inspection missions (especially in energy-constrained situations), existing technologies find it difficult to simultaneously take into account the requirements of global optimal path and energy consumption optimization. Summary of the Invention

[0005] In order to overcome the above-mentioned shortcomings, the present invention aims to provide an energy-optimized path planning method and system for drone inspections in complex environments, a corresponding drone and a computer-readable storage medium, and solve the problem of excessively high path costs in global path planning in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an energy-saving path planning method for drone inspection, the method comprising: Calculate the evaluation function when expanding a node at each step: in, Indicates the starting point to the current node The cumulative cost, heuristic function Represents a slave node Estimated cost to the target point, heuristic function By distance component and energy consumption constitute, Heuristic function The expression is: in, For the current node The straight-line distance to the target node, is the minimum additional energy consumption estimated based on altitude change and flight resistance, 、 is the weight coefficient.

[0007] The present invention also provides the following technical solutions: a UAV path planning system, comprising a global path planning module; The global path planning module is suitable for planning the global inspection path of the drone using the above-mentioned energy consumption optimization path planning method for drone inspection.

[0008] The present invention also provides the following technical solution: a patrol drone, comprising a processor and a memory arranged to store computer-executable instructions, wherein when the executable instructions are executed, the processor executes the above-mentioned energy-saving optimized path planning method for drone patrol.

[0009] The present invention also provides the following technical solution: a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by a processor, the energy-saving path planning method for drone inspection is implemented.

[0010] Compared with the existing technology, the beneficial effects of this technical solution are: During the global path planning phase, the present invention designs a multi-dimensional comprehensive heuristic function for UAV inspection energy consumption and flight constraints. This heuristic function incorporates an energy consumption model based on the traditional A* algorithm heuristic, taking into account factors such as Euclidean distance, terrain height difference, and flight resistance.

[0011] The improved heuristic function of the present invention can preferentially avoid high-energy consumption areas, select flatter and smoother routes, and significantly reduce the cumulative energy consumption and flight time of the global path; at the same time, the introduction of multidimensional costs makes global planning more accurate and reliable, and improves the search efficiency of the algorithm in complex three-dimensional environments.

[0012] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1 This is a schematic diagram of the main process of the path planning system of the present invention; Figure 2 It is the trigger judgment process for local dynamic obstacle avoidance; Figure 3 This is a schematic diagram of the fusion switching state of global planning and local planning; Figure 4 It is the structure diagram of the perception window and potential field guidance; Figure 5 Schematic diagram of the trajectory for avoiding obstacles in three-dimensional space. DETAILED DESCRIPTION

[0015] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0016] See also Figure 1-5 , a UAV path planning system, the overall architecture includes a global path planning module, a local obstacle avoidance planning module and a decision switching module. The system input is the coordinates of the starting point and target point of the UAV, environmental map information, and real-time perception data through the UAV's onboard sensors (for example, for detecting dynamic obstacles). The environmental map information may include map information such as the location of static obstacles, terrain height, flight restricted areas, strong wind areas, flight resistance, etc. The system output is a safe and feasible flight route (navigation point sequence) from the starting point to the target. The path is smooth and continuous and avoids dynamic obstacles. It can be used to guide the UAV's automatic cruising and inspection. The key components and workflow of the system of the present invention will be elaborated in detail below.

[0017] In this embodiment, in the global path planning module, at the start of a mission, if static map information of the environment is known, the present invention first uses an improved A* algorithm to perform global path planning. Global path planning is performed on an offline digital map or grid map, generating a preliminary optimal path from the starting point to the end point. Unlike the traditional A* algorithm, this invention introduces a multi-dimensional heuristic evaluation function to guide the search direction. The improved A* algorithm calculates the evaluation function at each node expansion step: in, Indicates the starting point to the current node The cumulative cost, heuristic function Indicates the current node The estimated cost to reach the destination node.

[0018] Heuristic function By distance component and energy consumption Composition, heuristic function The expression is: in, For nodes The straight-line distance to the target, is the minimum additional energy consumption estimated based on altitude change and flight resistance, 、 is the weight coefficient.

[0019] The present invention will The design is the aforementioned multi-dimensional comprehensive heuristic function to simultaneously consider factors such as distance and energy consumption to improve the rationality of global path evaluation.

[0020] For example, in flat airspace It places more emphasis on straight-line distance and increases the weight of energy consumption cost in complex terrain or strong wind areas, thereby adaptively balancing the two goals of shortest flight distance and lowest energy consumption.

[0021] When satisfied Consistency (i.e. the actual minimum cost from any node to the target is no less than ), the improved A* algorithm can still guarantee to find the global optimal solution. By adjusting the heuristic function weight, the global planning of the present invention not only ensures search efficiency, but also takes into account the economy and safety of drone flight.

[0022] In this embodiment, the output of global path planning is a navigation point sequence consisting of a series of ordered waypoints. ,in It's the starting point. is the target point, the others These navigation points are usually located in safe and open areas with few obstacles. They are the coordinate points that the drone should ideally pass through in sequence. The navigation point sequence provides a guiding skeleton for subsequent local planning: the drone will try to fly along this sequence in the direction of Past .

[0023] In some embodiments, if the environment map is unknown or incomplete, the present invention can operate in a heuristic exploration mode: Initially, global A* planning is skipped, and the system directly enters the local planning phase, gradually approaching the target through continuous perception and local search. During this process, the system gradually builds an environment map in the background, triggering global path planning as appropriate. Once a reliable global navigation point sequence is obtained, the drone can be guided according to the above process. In the following examples, it is assumed that the environment is known and the global navigation point sequence has been obtained to illustrate the system's operating logic under normal circumstances.

[0024] In this embodiment, after takeoff, the drone will fly along a path according to the sequence of navigation points provided by the global plan. During flight, the local obstacle avoidance planning module runs continuously to cope with sudden obstacles and environmental changes that may occur along the way. Local planning is based on the artificial potential field (APF) method for dynamic obstacle avoidance. The core concept is to regard the drone as a particle in motion under force. Targets in the environment generate a gravitational field on it, and obstacles generate a repulsive field. The drone adjusts its movement in the direction of the resultant force, thereby achieving obstacle avoidance and navigation. This invention has made multiple improvements to the classic APF algorithm, making it more suitable for the real-time obstacle avoidance needs of drones.

[0025] The drone's onboard perception system (such as lidar, camera, or radar) scans the surrounding environment at a fixed frequency to obtain the location, distance, and movement information of obstacles within the local range at the current moment. The perception range is set as a circular window centered on the drone (such as Figure 4 (as shown in the figure), the radius R is determined by the effective detection range of the sensor. A local grid map, known as the potential field grid map, is then created within this sensing circle. The potential field grid map discretizes the local space into a grid with a certain resolution, with each grid recording the potential energy value at that location. During initialization, the potential energy of each grid is set to 0. Then, based on the currently detected target points and obstacles, the potential field grid map is used to calculate the superposition of attractive and repulsive forces.

[0026] Specifically, first determine the local sub-goal set within the perception window: traverse the global navigation point sequence , select the navigation points that fall within the current perception range (radius R) and record them as a set These points are the temporary goals of this round of local planning. If the set G is not empty, each sub-goal point is Calculate the gravitational potential field: for example, using the quadratic gravitational potential energy function in, Indicates the current position of the drone. For drone to sub-target distance, is the gravitational coefficient. When the UAV is located in different grids, Calculate the gravitational vector based on the relative displacement The direction of the gravitational vector points to the sub-target , its size is proportional to the distance, that is, the farther the distance, the greater the gravity. In the case of multiple sub-targets, the direction of gravity generated by each target on the drone may be different. The present invention adopts the gravity synthesis strategy: all The resulting gravitational vectors are linearly superimposed to obtain the total gravitational force on the drone. This combined gravity simultaneously points toward a "balance of forces" at multiple target points, guiding the drone in the optimal direction. As can be seen, the combined gravity actually gives closer navigation points a greater pulling effect, rather than being restricted to a single fixed sub-target. This ensures that the drone receives continuous multi-target guidance: even if a navigation point is temporarily inaccessible due to an obstacle, the combined gravity will guide the drone toward other nearby navigation points, preventing it from becoming stagnant.

[0027] Next, based on the currently perceived obstacle set Construct a repulsive potential field. For each obstacle , first calculate the distance between the drone and it To ensure safety, the obstacle is considered as a slightly expanded entity when calculating the repulsive force, that is, the effective distance is used. ( is the effective radius of the obstacle) to provide the necessary safety margin. Less than the preset repulsive influence threshold When the obstacle generates repulsive force on the drone; when it exceeds The influence is ignored. Repulsion function in, is the repulsion coefficient. When, regulations From this potential energy, we can get the repulsive force vector: The direction of Decision, point away from obstacles The direction of the obstacle increases sharply as the distance decreases. When there are multiple obstacles, The total repulsive force is obtained by summing The dynamic obstacle prediction repulsion model of the present invention further introduces obstacle motion prediction on the basis of the above: for dynamic obstacles with speed, the preview time is selected , predicting that After the position ,in is the velocity vector of the obstacle, and the distance from the drone to the predicted position Instead of static distance , which is used to calculate the repulsive force. The predicted repulsive force of the dynamic obstacle will increase earlier than that of the static situation, so that the UAV can bypass it in advance, such as Figure 2 As shown, the local obstacle avoidance decision-making process first determines whether an obstacle is ahead of the route and calculates the potential collision risk based on the predicted trajectory before deciding whether to trigger obstacle avoidance. The dynamic repulsion model effectively prevents the drone from chasing moving targets or being caught off guard by sudden obstacles, improving the proactiveness and stability of obstacle avoidance.

[0028] The gravity of all the above sub-targets and the repulsion of obstacles are superimposed to form the resultant force currently acting on the drone. At this point, the problem of local planning of the drone is transformed into choosing the next movement position along the direction of the resultant force. In the continuous space, the drone will adjust its heading in real time to make the flight direction consistent with the direction of the resultant force. In the discrete grid diagram, the direction of the fastest potential energy decrease in the adjacent grid is selected (this is equivalent to sampling along the direction of the resultant force). The potential field grid diagram provides an intuitive basis for this process: the grid potential energy value is given by and Together, the direction of the gradient descent of the potential energy around the drone's current position is the local optimal direction of travel. The drone's local obstacle avoidance planning in this invention can be understood as a rolling optimization search on a potential field grid: each cycle, the potential field distribution is updated based on the latest environment, and then a step is moved toward a location with lower potential energy. This iterative process gradually approaches the local sub-target and avoids obstacles.

[0029] In this embodiment, after determining the direction of movement based on the combined force, the system can also appropriately adjust the speed and trajectory in combination with the kinematics of the drone to achieve smooth obstacle avoidance.

[0030] First, for the heading control of the drone, the angle between the resultant force direction and the current nose direction (for example, confirmed by the gyroscope, accelerometer, magnetometer, etc. on the drone) can be used as the steering angle that needs to be corrected. ,The UAV uses a track controller (such as heading angle PID or attitude solver) to gradually make the actual flight direction approach the direction of the resultant force, thus avoiding instability caused by sharp turns.

[0031] Secondly, in terms of speed control, when the distance to the obstacle ahead is close or the repulsive force is large, the flight speed should be appropriately reduced; on the contrary, when there is no obstruction, the cruising speed can be restored. This speed adjustment can be calculated linearly or nonlinearly based on the distance to the nearest obstacle: for example in, is the preset cruising speed, is the safety distance threshold. When it approaches 0, it will slow down to a hovering state. Through the above-mentioned speed-obstacle avoidance linkage control, the UAV can maintain a reasonable speed during the obstacle avoidance process, so as not to miss the turn due to excessive inertia or oscillate due to excessive avoidance.

[0032] The present invention can effectively solve the local minimum problem in the traditional artificial potential field method, and the situation where the drone is symmetrically surrounded by multiple obstacles causes the resultant force to be offset. , the drone may become stagnant or wander back and forth near a certain point. To this end, the present invention introduces a simulated annealing perturbation mechanism in local planning: when it is detected that the drone has failed to significantly approach the sub-target after multiple iterations or the speed has dropped to near zero, the simulated annealing algorithm is triggered to apply a random perturbation displacement or additional virtual force to the current position of the drone, causing it to temporarily deviate from its original equilibrium position. This disturbance has a large initial amplitude, guiding the drone out of the local trap, and then gradually reduces the disturbance amplitude to ensure that the movement tends to be stable. The simulated annealing mechanism has a certain degree of random exploration, which can effectively help the drone escape from the local minimum point and move in the right direction again. Combined with the above-mentioned multi-target gravitational mechanism and dynamic repulsion model, the local obstacle avoidance algorithm of the present invention can maintain reliable obstacle avoidance performance in most complex scenarios.

[0033] In this embodiment, the path planning system of the present invention realizes the organic integration of global planning and local planning through the decision switching module.

[0034] During the entire process of the UAV's mission, the system intelligently switches between the "driving along the global path" and "local obstacle avoidance and detour" modes based on real-time environmental analysis (e.g. Figure 3 Initially, the drone follows the global navigation point sequence. When the local obstacle avoidance module detects an obstacle ahead that may conflict with the current route (refer to Figure 2), the system immediately suspends strict global path following and instead executes local obstacle avoidance planning (APF algorithm) to circumvent the obstacle. Obstacle avoidance mode persists throughout the drone's circumvention of the obstacle: during this period, the drone relies primarily on the aforementioned potential field for guidance, with the original global target point retained only as a remote reference. Obstacle avoidance is determined to be complete based on the following conditions: the nearest obstacle has left the perception range or the drone has moved outside the obstacle's shadow on the global path (i.e., returning to an unobstructed area of ​​the originally planned path). Once obstacle avoidance is complete, the system switches back to global path following mode, guiding the drone toward the next global navigation point.

[0035] It's important to emphasize that by the time local obstacle avoidance concludes, the drone is often no longer on the precise trajectory of the originally planned path. If the deviation is small, no additional processing is required; the drone will fly directly toward the navigation point, gradually converging back to the global route. However, if the deviation is too large, the remaining path may become suboptimal or unconnected. To this end, the present invention incorporates a path deviation monitoring mechanism: it calculates the shortest distance from the current drone's position to the global path or the distance to the next navigation point. If this threshold is exceeded, the global path validity is considered to have decreased, triggering a new round of global path replanning. At this point, using the drone's current position as the new starting point and the original target point as the end point, an improved A* algorithm is run based on updated map information to generate a new globally optimal path (bypassing previously inaccessible areas), and the subsequent navigation point sequence is updated accordingly. The results of the global replanning replace the original navigation points that have not yet been reached, ensuring that the plan remains synchronized with environmental changes. If replanning is not triggered, the drone continues to travel along the original navigation point sequence.

[0036] In order to maintain a smooth transition of the trajectory during the mode switching process, the present invention introduces a dynamic balancing strategy of weight factors in the decision switching module. The influence of global path guidance and local potential field guidance on UAV motion is weighted separately: when there is no obstacle interference in the environment, At this time, the UAV strictly follows the globally planned route to fly; when an obstacle appears and needs to be avoided locally, the speed is appropriately reduced. The value is used to enhance the local guidance effect of APF and apply a moderate offset to the heading. The value is selected to achieve a dynamic balance and seamless connection between global planning and local obstacle avoidance guidance. This ensures that the drone flies along the optimal global route under normal circumstances, and can smoothly deviate from the original route to avoid obstacles when circumventing local obstacles. After the obstacle avoidance is completed, it can naturally transition back to the global route, avoiding sudden changes in the track during the switching process, reducing the computational overhead caused by frequent replanning, and ensuring the consistency and safety of the route in dynamic environments.

[0037] In this embodiment, the global / local switching process is repeated and continued until the UAV finally reaches the target point. Figure 1 As shown, the main process of the planning system of the present invention can be summarized as follows: initialization → global planning → loop: obstacle detection → local obstacle avoidance → path restoration → end of target detection. This architecture enhances global planning through event-driven local planning: the local planning module is enabled only when obstacle avoidance is required, which greatly reduces unnecessary computing overhead; after obstacle avoidance is completed, it immediately returns to the global path following to ensure the overall optimization of the route. With the help of multi-target guidance and path fusion strategies, the global and local planning modes are naturally connected during the transition, and no obvious path segmentation or oscillation occurs. For example, under the fusion strategy of the present invention, even if the drone frequently enters and exits the obstacle avoidance mode, its motion trajectory remains smooth and coherent, which is more in line with human driving intuition. Figure 3 Shown is a state transition diagram of the global-local planning fusion of the present invention.

[0038] In this embodiment, an energy-optimized path planning method for drone inspection is actually proposed. The improved A* algorithm mentioned above is used to calculate the evaluation function at each node expansion step: in, Indicates the starting point to the current node The cumulative cost, heuristic function Indicates the current node The estimated cost to reach the destination node.

[0039] Heuristic function By distance component and energy consumption Composition, heuristic function The expression is: in, For nodes The straight-line distance to the target, is the minimum additional energy consumption estimated based on altitude change and flight resistance, 、 is the weight coefficient.

[0040] In this embodiment, in fact, a patrol drone is also proposed, including a processor and a memory arranged to store computer-executable instructions. When the executable instructions are executed, the processor executes the energy-saving optimized path planning method for drone inspection mentioned above.

[0041] In this embodiment, in fact, a computer-readable storage medium is also proposed, which stores one or more programs. When the one or more programs are executed by the processor, the energy-saving and optimized path planning method for drone inspection mentioned above is implemented.

[0042] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. An energy-saving path planning method for drone inspection, characterized in that: Methods include: Calculate the evaluation function when expanding a node at each step: in, Indicates the starting point to the current node The cumulative cost, heuristic function Represents a slave node Estimated cost to the target point, heuristic function By distance component and energy consumption constitute, Heuristic function The expression is: in, For the current node The straight-line distance to the target node, is the minimum additional energy consumption estimated based on altitude change and flight resistance, 、 is the weight coefficient.

2. The energy consumption optimization path planning method for drone inspection according to claim 1 is characterized in that: The output of global path planning is a sequence of navigation points consisting of a series of ordered waypoints. ,in It's the starting point. is the target point, the others The target navigation point.

3. The energy consumption optimization path planning method for drone inspection according to claim 1 or 2 is characterized in that: Global path planning is performed on an offline digital map or raster map.

4. A UAV path planning system, characterized in that: Including global path planning module; The global path planning module is suitable for planning the global inspection path of the drone using the energy consumption optimization path planning method for drone inspection described in any one of claims 1-3.

5. A patrol drone, characterized in that: The invention comprises a processor and a memory arranged to store computer-executable instructions, wherein when the executable instructions are executed, the processor executes the energy-saving path planning method for unmanned aerial vehicle inspection according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores one or more programs, which, when executed by a processor, implement the energy-saving path planning method for drone inspection according to any one of claims 1 to 3.

Images