An unmanned aerial vehicle path planning method based on an improved ant colony algorithm

Abstract

The present application relates to the field of unmanned aerial vehicle path planning, in particular to an unmanned aerial vehicle path planning method based on improved ant colony algorithm. The present application proposes an improved ant colony algorithm. Firstly, the algorithm adds the evaluation function of A* algorithm in the heuristic function, and introduces the anti-bending weight coefficient to reduce the number of path bending and the large cumulative turning angle; secondly, the information pheromone deployment factor is used to improve the updating rule of information pheromone, and the convergence speed and global optimality of the algorithm are improved; finally, the backtracking mechanism and the information pheromone concentration punishment mechanism are introduced to solve the problem that the traditional ant colony algorithm is easy to fall into deadlock. The simulation results show that, compared with the traditional ant colony algorithm, the improved ant colony algorithm reduces the corner number by 67% and the iteration number by 75% in a more complex map. The improved ant colony algorithm has faster convergence speed and higher efficiency, and has more practical guiding significance for unmanned aerial vehicle path planning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle path planning, and particularly relates to an unmanned aerial vehicle path planning method based on an improved ant colony algorithm. BACKGROUND

[0002] Unmanned aerial vehicles have been widely used in important fields such as smart agriculture, industrial production, and logistics delivery, and have broad development prospects. In the field of unmanned aerial vehicle technology, path planning has always been an important research hotspot. The purpose of path planning is to quickly and accurately find a path connecting the starting point and the target point in a map with various obstacles, and the path can accurately avoid all obstacles. According to the characteristics of the path planning algorithm, they can be divided into traditional path planning algorithms and intelligent path planning algorithms. Traditional path planning algorithms include APF algorithm, Dijkstra algorithm, A* algorithm, Bug algorithm and the like. However, such algorithms have problems such as low calculation efficiency and easy to fall into local minimum. When the search map is expanded, these limitations will be magnified. Intelligent path planning algorithms include neural network algorithm, genetic algorithm and ant colony algorithm, and the biggest feature of such algorithms is that they have special learning ability and randomness, and can cope with more complex map environment, and the solution is not unique.

[0003] Among them, the ant colony algorithm in the intelligent algorithm has strong map dynamic search ability and good robustness, and is widely used in the field of path planning algorithm. However, it cannot be denied that the traditional ant colony algorithm also has its limitations, such as slow convergence speed, local optimum and easy to fall into deadlock. In the field of unmanned aerial vehicle path planning, few studies consider the problem of long running time and increased energy consumption caused by too many turning times and too large cumulative turning angles. SUMMARY

[0004] The purpose of the present application is to provide an unmanned aerial vehicle path planning method based on an improved ant colony algorithm, which aims to solve the problem that the traditional ant colony algorithm is easy to fall into deadlock and reduce the number of path bending times and the large cumulative turning angle of the unmanned aerial vehicle during planning.

[0005] To achieve the above purpose, the present application provides an unmanned aerial vehicle path planning method based on an improved ant colony algorithm, comprising the following steps:

[0006] Step 1: model the map environment by applying the grid method, and set the positions of the starting point and the target point;

[0007] Step 2: initialize the system and set the parameters;

[0008] Step 3: update the taboo list;

[0009] Step 4: judge the process deadlock;

[0010] Step 5: selecting the next grid by roulette strategy;

[0011] Step 6: judging whether the ant reaches the target point;

[0012] Step 7: updating pheromone.

[0013] Optionally, in the process of initializing the system and setting parameters, the corresponding parameters include the number of ants m, pheromone excitation factor a, expected heuristic factor b and pheromone evaporation coefficient p.

[0014] Optionally, the process of updating the tabu list is that the ant k (k = 1, 2,..., n) is placed on the current node, and the current node is added to the tabu list.

[0015] Optionally, the judging process of deadlock is that whether the ant falls into a deadlock point is judged, if it falls into a deadlock point, a backtracking mechanism is taken, and the node before backtracking is included in the tabu list, at the same time, the pheromone concentration punishment mechanism empties the pheromone concentration on the deadlock point, then whether the next grid is feasible and whether the target point is reached are judged, if the next grid is feasible and the target point is not reached, step 5 is turned to, otherwise step 3 is turned to.

[0016] Optionally, in the process of selecting the next grid by roulette strategy, the heuristic function is improved by adding A* algorithm and considering the target point, and then the next grid to be reached is selected.

[0017] Optionally, the process of judging whether the ant reaches the target point is that whether every ant has reached is judged, if not, step 3 is repeatedly executed until every ant in the ant colony completes path search in the iteration process, and then step 7 is turned to.

[0018] Optionally, the process of updating pheromone includes the following steps:

[0019] Step 7.1: after each iteration, if the number of iterations meets the requirement, the pheromone is updated according to the following formula:

[0020] τ ij (t+1) = (1-p)τ ij (t) + Δτ ij

[0021]

[0022]

[0023]

[0024]

[0025] W2=V uav ·T uav

[0026] Wherein, the value range of p is 0 < p < 1, K1 and K2 are constants greater than 1, indicates pheromone deployment factor, is a variable constant, L k Indicates the path length searched by the ant k, Indicates the optimal path turning weight, L average Indicates the average path length of all ants in this iteration, W1 and W2 respectively indicate the conversion of turning angle and turning number into grid length, Cals(l) indicates the sum of all angles of turning on the optimal path, Turn(l) indicates the sum of turning numbers on the optimal path, V uav Indicates the constant speed of the unmanned aerial vehicle flight, omega uav Indicates the angular velocity of the unmanned aerial vehicle turning flight, T uav Indicates the time of accelerating or decelerating once for the unmanned aerial vehicle turning;

[0027] Step 7.2: Ensure that the pheromone concentration meets the following formula:

[0028]

[0029] Wherein, tau max And tau min Respectively, the upper and lower limits of the pheromone concentration set on each path; if the convergence condition is met, exit; if not, go to step 3;

[0030] Step 7.3: If the iteration number N meets N > N max , stop counting, and output the final result.

[0031] The application provides a kind of unmanned aerial vehicle path planning method based on improved ant colony algorithm, in heuristic function, while introducing anti-bending weight coefficient, solve path bending number and the problem of larger cumulative turning angle, the evaluation function of improved A* algorithm is introduced;Then the updating rule of pheromone is improved by using pheromone deployment factor, to improve the convergence speed and global optimality of algorithm;Finally, introduce back mechanism and pheromone concentration punishment mechanism to solve the problem that traditional ant colony algorithm is easy to fall into deadlock.The application can reduce the path bending number and the larger cumulative turning angle of unmanned aerial vehicle in planning, improve the convergence speed and global optimality of algorithm, and solve the problem that traditional ant colony algorithm is easy to fall into deadlock.From the simulation effect, the improved ant colony algorithm of the application converges faster, has fewer corner numbers, is more efficient, and has more realistic guiding significance for unmanned aerial vehicle path planning. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0033] Figure 1 is a specific flowchart of a UAV path planning method based on an improved ant colony algorithm of the present application.

[0034] Figure 2 is a moving track schematic diagram of a traditional ant colony algorithm in a 10*10 map in a specific embodiment of the present application.

[0035] Figure 3 is a moving track schematic diagram of an improved ant colony algorithm in a 10*10 map in a specific embodiment of the present application.

[0036] Figure 4 is a convergence curve schematic diagram of a traditional ant colony algorithm in a 10*10 map in a specific embodiment of the present application.

[0037] Figure 5 is a convergence curve schematic diagram of an improved ant colony algorithm in a 10*10 map in a specific embodiment of the present application.

[0038] Figure 6 is a moving track schematic diagram of a traditional ant colony algorithm in a 20*20 map in a specific embodiment of the present application.

[0039] Figure 7 is a moving track schematic diagram of an improved ant colony algorithm in a 20*20 map in a specific embodiment of the present application.

[0040] Figure 8 is a convergence curve schematic diagram of a traditional ant colony algorithm in a 20*20 map in a specific embodiment of the present application.

[0041] Figure 9 is a convergence curve schematic diagram of an improved ant colony algorithm in a 20*20 map in a specific embodiment of the present application. DETAILED DESCRIPTION

[0042] The embodiments of the present application will be described in detail below, and examples of the embodiments are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0043] The present application provides a UAV path planning method based on an improved ant colony algorithm, comprising the following steps:

[0044] S1: model the map environment by applying the grid method, and set the positions of the starting point and the target point;

[0045] S2: initialize the system and set parameters;

[0046] S3: update the tabu list;

[0047] S4: judge whether the process is deadlocked;

[0048] S5: select the next grid by adopting the roulette strategy;

[0049] S6: judge whether the ant has reached the target point;

[0050] S7: update the pheromone.

[0051] The specific overall flowchart is shown in Figure 1 The following is further explained in combination with specific implementation steps and related terms:

[0052] 1. State transition probability

[0053] At time t, the next path node j is selected by the current node i, which is determined by the ant k according to the initial pheromone concentration of the system, the heuristic factor and other factors, and the roulette strategy is adopted, so the state transition probability of the next node is:

[0054]

[0055] In the formula, τ ij represents the pheromone concentration from grid i to grid j, η ij represents the heuristic information from grid i to grid j, alpha represents the pheromone excitation factor, the greater the value of alpha, the greater the probability that the path walked by most ants is selected. Beta is the expected heuristic factor, which represents the relative influence degree of the heuristic information. allowed k is the set of nodes available for selection by the ant k.

[0056] 2. Heuristic function

[0057] The traditional ant colony algorithm only considers the cost between node i and node j when searching for a path, and does not take the influence of the target point into account, which leads to a small difference in the heuristic value between adjacent grids in the map, and also leads to the algorithm being easily trapped in a local optimum during path searching, which is not conducive to the global optimal solution. The A* algorithm has a better planning path and planning speed when planning globally. Based on this, the A* algorithm is added to the heuristic function of the traditional ant colony algorithm, and the target point is taken into account, and the evaluation function f(n) of the A* algorithm is used to guide the search and expansion of the next node by the ant. The expression of the evaluation function f(n) is shown in formula (2):

[0058] f(n)=g(n)+h(n) (2)

[0059] Wherein, g(n) represents the minimum path cost value from the starting node to the current node, h(n) represents the minimum path cost estimation value from the current node to the target node.The expression of g(n) and h(n) is shown as formula (3)~(4):

[0060]

[0061]

[0062] Wherein, (n x ,n y ) represents the coordinates of the current node n, (s x ,s y ) represents the coordinates of the starting node s, (g x ,g y ) represents the coordinates of the target node g.

[0063] If the UAV only considers the shortest path in actual path planning, the problem of long running time and increased energy consumption caused by too many turning times and too large cumulative turning angle will be caused.Based on this, the anti-bending weight coefficient ξ bend Is introduced to make the UAV select a path with less turning times and small turning angle as much as possible, and the anti-bending weight coefficient ξ bend The expression is shown as formula (5):

[0064]

[0065] Wherein, Is the coefficient of converting path bending times into grid length, and ψ is the coefficient of converting angle into grid length.turn path Represents the path bending times of the UAV from the previous node n-1 to the next node n+1, and angle path Represents the angle between the line segment from the previous node n-1 to the current node n and the line segment from the current node n to the next node n+1.

[0066] In summary, in the improved ant colony algorithm, the evaluation function of A* algorithm is added in the heuristic function, which can improve the global search ability of the algorithm and improve the convergence speed;At the same time, the anti-bending weight coefficient is introduced in order to reduce the path bending times and the large cumulative turning angle.The improved heuristic function is shown as formula (6):

[0067]

[0068] Wherein, C1 is a constant greater than 1, and λ1 and λ2 represent heuristic stimulation factors.

[0069] 3. Pheromone update rule

[0070] The update rule of the pheromone concentration of the traditional ant colony algorithm is mainly composed of the remaining part after volatilization with time and the newly added part after the next round of iteration ants, which may cause the problems of low search efficiency and slow convergence speed; and when searching the path, a part of the ants search the worst path, and the pheromone released on the worst path has a negative effect on the ants in the later iteration, which is easy to fall into a local optimal solution; and another part of the ants search the optimal path, and the pheromone released on the optimal path has a positive feedback effect on the ants in the later iteration. Based on this, the global optimality is comprehensively considered, the pheromone update rule is improved, the pheromone left on the path with better performance is larger, the pheromone left on the path with worse performance is smaller, and the convergence of the algorithm is improved. The improved pheromone update rule is shown in formulas (7)-(8):

[0071] τ ij (t+1)=(1-ρ)τ ij (t)+Δτ ij (7)

[0072]

[0073] Wherein, the value range of p is 0 < p < 1, K1 and K2 are constants greater than 1, is an information pheromone allocation factor, is a variable constant, L k represents the path length searched by the ant k, represents the optimal path turning weight, and The expressions are shown in formulas (9)-(12):

[0074]

[0075]

[0076]

[0077] W2=V uav ·T uav (12)

[0078] Wherein, L average represents the average path length of all ants in this iteration, W1 and W2 respectively represent the conversion of the turning angle and the number of turns into the grid length, Cals(l) represents the sum of all angles of the turns on the optimal path, Turn(l) represents the sum of the number of turns on the optimal path, V uav represents the constant speed of the unmanned aerial vehicle flight, and uav represents the angular velocity of the unmanned aerial vehicle turning flight, and uavRepresents the time of acceleration or deceleration of the UAV for one rotation.

[0079] After the improved algorithm is iterated for several times, the pheromone concentration on some paths can be too large or too small, which makes the subsequent search difficult to proceed. In order to prevent the occurrence of such precocious conditions, with reference to the Max-Min Ant System (MMAS algorithm), the value of the pheromone concentration tau is limited, as shown in formula (13):

[0080]

[0081] Wherein, tau max and tau min are the upper and lower limits of the pheromone concentration set on each path.

[0082] 4. Deadlock problem

[0083] When the map environment is relatively complex, such as there is a U-shaped obstacle, the ants can be trapped in a deadlock point when searching for a path and cannot continue to move to the next node. When the traditional ant colony algorithm is trapped in such a situation, the ants trapped in the deadlock point are selected to be removed from the ant colony, and the global pheromone is not updated. However, such a defect is that if the deadlock point is not limited, a part of the ants will still be trapped in the deadlock point, and after several rounds of iteration, the number of ants in the ant colony will become less and less, which reduces the diversity of the global solution and is not conducive to the search of the optimal path. Based on this, the present application proposes a back mechanism and a deadlock point pheromone concentration punishment mechanism, that is, when the ant k is trapped in the deadlock point, the back mechanism allows the ant k to back up one step, and the node before backing up is included in the taboo table, and the pheromone concentration punishment mechanism empties the pheromone concentration on the deadlock point, that is, the pheromone is 0. If the ant is still trapped in the deadlock state, the above steps are repeated until it is no longer trapped in the deadlock state.

[0084] Further, in step S2, the relevant parameters include the number of ants m, the pheromone incentive factor alpha, the expected heuristic factor beta, and the pheromone evaporation coefficient rho.

[0085] The process of updating the taboo table in step S3 is to place the ant k (k = 1, 2,..., n) on the current node and add the current node to the taboo table.

[0086] In step S4, it is judged whether the ant is trapped in the deadlock point. If it is trapped in the deadlock point, the back mechanism is adopted, and the node before backing up is included in the taboo table, and the pheromone concentration punishment mechanism empties the pheromone concentration on the deadlock point. Then it is judged whether the next grid of the ant is feasible and whether the target point has been reached. If the next grid is feasible and the target point has not been reached, go to step S5, otherwise go to step S3.

[0087] Step S5, the next grid is selected, the heuristic function is calculated according to formula (6) and the probability function is calculated according to formula (2), and then the next grid to be reached is selected by using a roulette strategy.

[0088] Step S6: it is judged whether the ant reaches the target point, if not, step S3 is repeated until each ant in the ant colony completes path search in the iteration process, and then step S7 is turned to.

[0089] Step S7: pheromone is updated. After each iteration, if the iteration number N satisfies N≤N max , the pheromone is updated according to formulas (7)-(12), while ensuring that the pheromone concentration satisfies formula (13), if the convergence condition is satisfied, it is exited; if not, it is turned to step 3; if the iteration number N satisfies N>N max , the counting is stopped, and the final result is output.

[0090] Further, please refer to Figures 2 to 9 , the application also proposes a specific embodiment, which is compared with the traditional ant colony algorithm by simulation results:

[0091] 1) 10*10 map

[0092] The path moving trajectories of the traditional ant colony algorithm and the improved ant colony algorithm of the application in the 10*10 map are as shown in Figure 2 and Figure 3 . As shown in the figures, in the 10*10 grid map, the improved algorithm has certain superiority, the corner number of the traditional ant colony algorithm is 7 times, while the corner number of the improved ant colony algorithm is 4 times, and the corner number is reduced by 43%. The convergence curves of the traditional ant colony algorithm and the improved ant colony algorithm of the application are as shown in Figure 4 and Figure 5 . As shown in the figures, the optimal path iteration number of the traditional ant colony algorithm is 9 times, while the optimal path iteration number of the improved ant colony algorithm is 4 times, and the optimal path iteration number is reduced by 56%.

[0093] 2) 20*20 map

[0094] The path moving trajectories of the traditional ant colony algorithm and the improved ant colony algorithm of the application in the 20*20 map are as shown in Figure 6 and Figure 7 . As shown in the figures, in the 20*20 grid map, the effect of the improved ant colony algorithm of the application is more obvious, the generated path is smoother, the corner number of the traditional ant colony algorithm is 15 times, while the corner number of the improved ant colony algorithm is only 5 times, and the corner number is reduced by 67%, which avoids excessive energy loss of the unmanned aerial vehicle and prolongs the service life of the unmanned aerial vehicle. The convergence curves of the traditional ant colony algorithm and the improved ant colony algorithm of the application are as shown in Figure 8 and Figure 9As shown in the figure, the improved ant colony algorithm has better overall optimization effect and faster convergence speed. The optimal path of the traditional ant colony algorithm converges after 63 iterations, while the optimal path of the improved ant colony algorithm converges after only 16 iterations, which reduces the iteration number by 75%.

[0095] In summary, the improved ant colony algorithm has certain advantages in a 10*10 grid map. Compared with the traditional ant colony algorithm, the number of corners is reduced by 43%, and the number of optimal path iterations is reduced by 56%. In a 20*20 complex grid map, the improved ant colony algorithm has more obvious effect, and the generated path is smoother. Compared with the traditional ant colony algorithm, the number of corners is reduced by 67%, which avoids excessive energy loss of the unmanned aerial vehicle and prolongs the service life of the unmanned aerial vehicle. The improved ant colony algorithm has only 16 iterations to reach the convergence state, which reduces the iteration number by 75% compared with the traditional ant colony algorithm. The improved ant colony algorithm has better overall optimization effect, faster convergence speed, fewer path turning numbers and smaller cumulative turning angles.

[0096] The above disclosure is only one preferred embodiment of the present application, and of course cannot limit the scope of the rights of the present application. Those skilled in the art can understand that the above-mentioned embodiment can be implemented in whole or in part, and equivalent changes made according to the claims of the present application still belong to the scope covered by the present application.

Claims

1. An unmanned aerial vehicle path planning method based on an improved ant colony algorithm, characterized in that, Comprise the following steps: Step 1: the grid method is applied to model the map environment, and the positions of the starting point and the target point are set; Step 2: initialize the system and set parameters; Step 3: update the taboo list; Step 4: judge the process deadlock; Step 5: select the next grid by using the roulette strategy; Step 6: judge whether the ant reaches the target point; Step 7: update the pheromone; The process of updating the pheromone comprises the following steps: Step 7.1: after each iteration, if the iteration number meets the requirement, update the pheromone according to the following formula: τ ij (t+1) = (1 - p)τ ij (t) + Δτ ij W2 = V uav • T uav where τ ij represents the pheromone concentration from grid i to grid j, ρ is in the range of 0 < ρ < 1, K1 and K2 are constants greater than 1, represents the pheromone deployment factor, which is a variable constant, L k represents the path length searched by ant k, represents the optimal path turning weight, L average represents the average path length of all ants in this iteration, W1 and W2 represent the conversion of turning angle and turning number into grid length, Cals(l) represents the sum of all angles of turning on the optimal path, Turn(l) represents the sum of the number of turns on the optimal path, V uav represents the constant speed of the UAV flight, ω uav represents the angular velocity of the UAV when turning, T uav represents the time for the UAV to accelerate or decelerate once when turning. Step 7.2: ensure that the pheromone concentration meets the following formula: wherein τ max and τ min are the upper and lower limits of pheromone concentration set on each path; if the convergence condition is met, exit; if not, go to step 3; Step 7.3: If the iteration number N satisfies N > N max then stop counting and output the final result.

2. The unmanned aerial vehicle path planning method based on the improved ant colony algorithm according to claim 1, wherein, In the process of initializing the system and setting parameters, the corresponding parameters include the number of ants m, the pheromone incentive factor a, the expected heuristic factor b, and the pheromone evaporation coefficient p.

3. The unmanned aerial vehicle path planning method based on the improved ant colony algorithm according to claim 2, wherein, The process of updating the taboo list is to place the ant k (k = 1, 2,..., n) on the current node and add the current node to the taboo list.

4. The unmanned aerial vehicle path planning method based on the improved ant colony algorithm according to claim 3, wherein, The process of judging the process deadlock is to judge whether the ant falls into a deadlock point, if it falls into a deadlock point, a backtracking mechanism is adopted, and the node before backtracking is included in the taboo list, and the pheromone concentration punishment mechanism empties the pheromone concentration on the deadlock point, then it is judged whether the next grid is feasible and whether the target point is reached, if the next grid is feasible and the target point is not reached, step 5 is turned to, otherwise step 3 is turned to.

5. The unmanned aerial vehicle path planning method based on the improved ant colony algorithm according to claim 4, wherein, In the process of selecting the next grid by using the roulette strategy, the A* algorithm is added and the target point is considered to improve the heuristic function, and then the next grid to be reached is selected.

6. The unmanned aerial vehicle path planning method based on the improved ant colony algorithm according to claim 5, wherein, The process of judging whether the ant reaches the target point is to judge whether each ant has reached, if not, step 3 is repeated until each ant in the ant colony completes the path search in the iteration process, and then step 7 is turned to.

Images