Intelligent unmanned aerial vehicle path planning system

Abstract

The invention discloses an intelligent unmanned aerial vehicle path planning system based on a reverse hyperbolic artificial bee colony algorithm, and aims to solve the problem of unmanned aerial vehicle path planning through an innovative algorithm. According to the system, a terrain environment model is established, and unmanned aerial vehicle path planning is carried out between two demand points needing to be planned. And efficient and safe path planning is realized while the generated path is ensured to comply with the flight limitation of the unmanned aerial vehicle. A preliminary test is carried out on a plurality of terrain models, and a result shows that the system can plan a smoother path with lower flight cost. Under the background of global economic acceleration and unmanned aerial vehicle application rapid growth, unmanned aerial vehicle path planning becomes a key link. The system comprehensively considers the terrain environment, fully searches the feasible space, improves the convergence and diversity of the algorithm, and finally generates a reliable path. Through the intelligent unmanned aerial vehicle path planning system based on the reverse hyperbolic artificial bee colony algorithm, the unmanned aerial vehicle can complete tasks more quickly and efficiently, fuel oil is saved, and overall benefit maximization is achieved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent optimization, aims to meet the actual needs of drone path planning, and designs an intelligent drone path planning system. The system uses an innovative algorithm to solve the drone path planning model based on the established terrain environment model. The mechanical property constraints of the drone itself are fully considered, and the algorithm is targeted and innovated according to the drone path planning. Background Art

[0002] With the improvement of UAV manufacturing technology, the reduction of UAV production and application costs, and the rapid development of technologies such as electronics and sensors, robots, automation and artificial intelligence, various UAV-based applications are becoming more and more popular. As an important research field of artificial intelligence, UAVs are widely used in various fields due to their convenient operation and good maneuverability.

[0003] There are many types of drones, which can be mainly divided into fixed-wing drones, multi-rotor drones, etc. according to their different structures. Among them, fixed-wing drones have the characteristics of fast flight speed, high mission execution efficiency and long endurance, but are not suitable for tasks such as staying in the air; multi-rotor drones have the characteristics of simple structure, convenient maneuverability and strong operability. With the improvement of informatization, people's requirements for the autonomous flight and group operation capabilities of intelligent drones are constantly increasing. In this context, path planning technology, as a part of the drone intelligent system, has undoubtedly become a hot topic of current research, which is related to the safety of drone flight in complex environments and the efficiency of mission execution.

[0004] Compared with traditional path planning methods, the intelligent UAV path planning system can customize the established terrain environment model and efficiently solve the path based on the model. The spatial environment of UAV path planning is complex and more environmental factors need to be considered. The current algorithm has limitations when facing more complex three-dimensional environments. For example, in an environment with more simulated mountains, the population converges slowly and is prone to falling into local minima. It is impossible to plan a short, low-threat path, making it difficult to meet the needs of efficient and high-quality planning.

[0005] With the increasing development and application of drones, the flight environment of drones is becoming increasingly complex. When performing tasks, drones need to face some problems such as long flight paths, complex tasks, large drone losses, and obstacles that are difficult to avoid. Therefore, achieving efficient and accurate path planning and ensuring the safety of drones has become a research focus. The main goal of this system is efficient, accurate, and safe path planning. It comprehensively considers the mechanical property constraints of drones and solves the characteristics. Through the intelligent drone path planning system, according to the demand points that need to be solved and the terrain environment, the path planning is more accurate and efficient to maximize the overall benefits. Summary of the invention

[0006] The implementation of the present invention is mainly divided into two parts: problem modeling and algorithm optimization.

[0007] 1. Problem Modeling

[0008] The UAV path planning problem needs to be based on the UAV path planning model and the path planning on the terrain environment model. The terrain environment model includes the use of geometric methods, that is, modeling the actual space through mathematical functions, which can intuitively describe the environment and has high environmental accuracy. The elevation map method is used to build a three-dimensional map with height information, and the ground elevation is obtained by using the height information in an ordered array. The system uses the geometric method to build two digital maps and the elevation map method to build a real map. The path planning model is established according to the limitations of the UAV itself. The model includes two objective functions of path planning (path length, path risk) and three UAV flight constraints (altitude constraint, horizontal flight angle and vertical flight angle).

[0009] 2. Algorithm optimization

[0010] UAV path planning involves constrained optimization problems, that is, objective optimization problems with multiple constraints. In multi-objective optimization problems, optimizing the performance of one objective may have a significant negative impact on the performance of other objectives. Traditional mathematical optimization algorithms often have difficulty dealing with multi-objective optimization problems where the objective function is not differentiable or has no clear mathematical expression. Evolutionary algorithms have a natural advantage in solving multi-objective optimization problems because they do not need to make any assumptions about the problem. Therefore, evolutionary algorithms are used to solve multi-objective optimization problems.

[0011] Unmanned aerial vehicle (UAV) path planning problem Unmanned aerial vehicles (UAV) are now widely used in the real world, such as search and rescue, crop dusting, monitoring, coverage planning and other fields. How to make the UAV fly to the destination without collision is one of the key issues in UAV technology, namely, UAV path planning. UAV path planning can be expressed as an optimization problem. The result is to find a path with the lowest cost while satisfying the constraints. This problem is a non-deterministic polynomial (NP) problem. Due to the complex spatial environment of UAV path planning, there are more environmental factors to consider. The current algorithm has limitations when facing a more complex three-dimensional environment. For example, in an environment with more simulated mountains, the population converges slowly and is prone to fall into the local minimum. It is impossible to plan a short and threatening path, which is difficult to meet the requirements of efficient and high-quality planning. It is a common method to use heuristic algorithms to solve the UAV path planning problem.

[0012] When using heuristic algorithms to solve problems, targeted improvements are required based on the fact that path planning is very likely to fall into local minima. A two-stage search strategy based on hyperbolic functions is used. In order to fully develop the search space, the development is divided into two stages. In the first development stage, i.e., the first half of the iteration, the space near the current solution is developed; in the second development stage, i.e., the second half of the iteration, in-depth development is carried out around the current optimal solution, and the intensity of development increases with the increase in iterations. This is a reverse learning search strategy, through which a refracted reverse solution is obtained. By comparing the fitness values ​​of the current solution and the refracted reverse solution, better individuals are selected to form a new population to participate in the next iteration. While ensuring population diversity, the algorithm is more likely to jump out of the local optimum and reach the global optimal area. The application of these methods makes problem solving more reliable and feasible, and more efficient.

[0013] Advantages of the present invention:

[0014] The UAV path planning based on the reverse hyperbolic artificial bee colony algorithm has significant advantages over traditional path planning methods. It largely solves the problem that the spatial environment of UAV path planning is complex and more environmental factors need to be considered. The current algorithm has limitations when facing a more complex three-dimensional environment. For example, in an environment with many simulated mountains, the population converges slowly and is prone to falling into a local minimum. The two-stage search strategy based on hyperbolic functions and the refractive reverse learning mechanism make the algorithm more convergent and diverse, and the UAV path obtained is shorter and safer, so that the UAV can complete the task more efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the terrain environment model. The red points represent the demand points.

[0016] Figure 2 It is a schematic diagram of the path planning modeling process.

[0017] Figure 3 It is a schematic diagram of the solution process. DETAILED DESCRIPTION

[0018] This invention is an intelligent UAV path planning system. The UAV path planning is optimized and solved to plan a short, smooth and safe UAV path. The mechanical constraints of the UAV itself, such as altitude constraints, horizontal flight angles and vertical flight angles, are considered, and the threat level of the planned path and terrain environment are considered. The model is modeled as a constrained optimization problem, and the reverse hyperbolic artificial bee colony algorithm is used to solve it to calculate a short, smooth and safe UAV path.

[0019] The detailed steps are as follows:

[0020] Step 1: Input Data

[0021] Convert the relevant data into .mat files in advance and read them into the matlab buffer area.

[0022] Step 2: Problem Modeling

[0023] In the problem modeling phase, we determined the variables and objective functions and constraints. The decision variables include the coordinates of the demand point and the terrain function or elevation digital map. Our goal is to plan a safe drone path with a short path length.

[0024] The objective function includes path length and path risk:

[0025] Path Length:

[0026] The path length consists of the distances between s+2 path points, dt k Indicates the length between the k+1th path points.

[0027]

[0028] Path risk:

[0029]

[0030] Where np is the number of the nearest grid points, dm i,j is the distance from the jth path point to the ith grid point. Introduce the safety distance dm safe To ensure that the drone avoids terrain obstacles, all grid points and path points of the terrain can be projected onto the horizontal plane, and then the safe distance dm can be obtained. safe The nearest grid points within range that may threaten the safety of the drone.

[0031] constraint:

[0032] Flight altitude constraints:

[0033]

[0034] Among them, (xp k ,yp k ,zp k ) is the path point p k The coordinate value, z min is the minimum safe height in the vertical direction, f(xp k ,yp k ) is a terrain function, which can calculate the z-axis coordinate value of the corresponding coordinate terrain.

[0035] Horizontal rotation constraint:

[0036]

[0037] Among them, p k ,p k+1 are two consecutive path points, and the value of k is between 0 and s.

[0038] Vertical slope constraint:

[0039]

[0040] Step 3: Optimization solution

[0041] Using the reverse hyperbolic artificial bee colony algorithm for optimization is an effective method. In the bee-following stage, the idea of ​​hyperbolic function is used for search. In order to fully develop the search space, the development is divided into two stages. In the first development stage, i.e. the first half of the iteration, the space near the current solution is developed; in the second development stage, i.e. the second half of the iteration, in-depth development will be carried out around the current optimal solution, and the development intensity increases with the increase of iterations.

[0042] Development formula for the first phase:

[0043]

[0044] W1=r1·a1·(cosh·r2+u·sinh·r2-1) (3.1)

[0045]

[0046] Where r is a random number in the interval [-1,1], r1, r2 are random numbers in the interval [0,1], k is selected by the roulette method in the original ABC, W1 is the weight coefficient of the first stage development, which controls the candidate solution to develop the space around itself from near to far. a1 is a monotonically decreasing function, t represents the current number of iterations, and T represents the maximum number of iterations. m, u are the sensitivity coefficients of control accuracy, according to, m = 0.45, u = 0.388.

[0047] Phase 2 development formula:

[0048]

[0049] W2=r5·a2 (3.1)

[0050]

[0051] r3, r4, r5 are random numbers in the interval [0, 1], W2 is used to control the development process of the second stage, controlling the candidate solution to fully search the surrounding space, a2 is a monotonically decreasing function. n is the sensitivity coefficient of the control accuracy, which is 0.5. By observation, the development of the original algorithm following the bee stage is very similar in structure to the modified formula (21) and formula (24).

[0052] The refraction reverse learning mechanism is used to obtain the refraction reverse solution. By comparing the fitness values ​​of the current solution and the refraction reverse solution, the better individuals are selected to form a new population to participate in the next iteration. While ensuring the diversity of the population, the algorithm is more likely to jump out of the local optimal area and reach the global optimal area.

[0053] The calculation formula of the refraction reverse learning solution is:

[0054]

[0055] Among them, x i,j represents the value of the i-th individual in the current population in the j-th dimension, is the reverse solution of refraction, min j ,max j Represents the lower and upper limits of the current population in the jth dimension. The perturbation vector φ is a random number uniformly distributed in the interval [0,h], which expands the search field, and h is the upper limit of the perturbation. k and n are related to the distance between the refraction and reflection solution and the local optimal solution. t represents the current number of iterations, and T represents the maximum number of iterations.

Claims

1. An intelligent UAV path planning system based on the reverse hyperbolic artificial bee colony algorithm, which solves the key problem that complex terrain environments are difficult to model, and path planning often falls into local optimality and cannot plan a smooth and low-cost path. It is characterized by: The system includes: terrain environment modeling module, path planning modeling module, and planning solution module; The terrain environment modeling module is used to establish a model of the UAV flight environment. The system can simulate digital terrain and real terrain; The path planning modeling module is used to establish a UAV path planning model and establish an optimization model based on the limitations of the UAV itself. The model includes two objective functions of path planning (path length, path risk), three UAV flight constraints (altitude constraint, horizontal flight angle and vertical flight angle; The planning and solving module is used to solve the path, solve the optimization model established in the path planning modeling module, and generate a smooth and low-cost path.

2. The intelligent UAV path planning system based on the reverse hyperbolic artificial bee colony algorithm according to claim 1 is characterized in that: The terrain environment modeling module constructs a smooth terrain environment and demand points for path planning according to a given terrain function or a real digital elevation model map.

3. The intelligent UAV path planning system based on the reverse hyperbolic artificial bee colony algorithm according to claim 1 is characterized in that: The path planning modeling module regards the path planning problem of UAVs between demand points in a terrain environment as a dual-objective optimization problem with three constraints, and considers two objective functions (path length, path risk) and three UAV flight constraints (altitude constraint, horizontal flight angle and vertical flight angle) in detail.

4. The intelligent UAV path planning system based on the reverse hyperbolic artificial bee colony algorithm according to claim 1 is characterized in that: The planning and solving module solves the established path planning model on the terrain environment model according to the reverse hyperbolic artificial bee colony algorithm, uses a two-stage search strategy based on hyperbolic functions in the algorithm search stage, and adopts the idea of ​​hyperbolic functions in the following bee stage. It is developed in two stages and uses the strategy of refractive reverse learning to generate a reverse population, improve population diversity, and generate a feasible, short-length, and low-flight-risk path.

Images