Unmanned vehicle obstacle avoidance path planning method

Abstract

The invention provides an unmanned vehicle obstacle avoidance path planning method, and relates to an unmanned vehicle path planning method, which comprises the following steps of: adding an improved gravitational potential field to a target point, adding an improved repulsive potential field to an obstacle, meanwhile, a road boundary potential field and a speed repulsion potential field are introduced to ensure that the unmanned vehicle does not collide with a road boundary and can avoid static and dynamic obstacles; when the path nodes are beyond the maximum influence distance of the obstacle, a traditional gravitational potential field is added to the two adjacent path nodes in the bat position updating formula, and the path length is shortened; in the updating process of the local new solution of the bat algorithm, two strategies of Gaussian distribution and Cauchy distribution are used for searching to generate the local new solution, so that the searching speed and the searching precision are improved. Based on the improvement, static and dynamic obstacle avoidance can be realized while the path is ensured to be as short as possible in the driving process of the unmanned vehicle, and a smooth path is generated.

Description

Technical Field

[0001] The present invention relates to a method for planning a path for an unmanned vehicle, and in particular to a method for planning an obstacle avoidance path for an unmanned vehicle. Background Art

[0002] The emergence of cars has brought great convenience to people's travel, but there are also some risks. The number of cars on the road continues to increase every year, and the frequency of accidents is also increasing year by year. At the same time, with the continuous development of science and technology, intelligence is combined with cars, that is, unmanned vehicles are developed. In unmanned driving technology, in order to meet people's travel needs, it is necessary to plan a safe and obstacle-free (static, dynamic) road in different scenarios to ensure safe and stable driving of the vehicle.

[0003] The bat algorithm is a heuristic intelligent algorithm based on the wisdom of biological groups. It can effectively search for the global optimal solution and has the advantages of simple model, fast convergence speed and good robustness. However, the traditional bat algorithm is prone to premature convergence and fall into local optimality in practical applications due to the insufficient initial search range. The convergence speed is slow in the later stage, which has a certain impact on the efficiency of the entire algorithm. The artificial potential field method (APF) was first proposed by Dr. Khatib in 1986. It is assumed that obstacles produce repulsive forces and target points produce gravitational forces. The path planning is carried out through the combined force potential field under the combined action of repulsive and gravitational fields, so that the vehicle moves in the direction of the potential field gradient descent, so as to reach the target point without collision. This method is simple, intuitive and widely used, but it is also prone to local optimal solution problems and unreachable target problems. Summary of the invention

[0004] The purpose of the present invention is to provide a method for obstacle avoidance path planning for unmanned vehicles. The method improves two traditional algorithms, the bat algorithm and the artificial potential field method, and then integrates the two algorithms to solve the obstacle avoidance path planning for unmanned vehicles. While ensuring that the path is as short as possible, static and dynamic obstacle avoidance is achieved to generate a smooth path.

[0005] The technical solution adopted by the present invention is:

[0006] A method for obstacle avoidance path planning for an unmanned vehicle, comprising the following steps:

[0007] Step S1, establish a map environment and initialize parameters;

[0008] Step S2, establishing an improved artificial potential field model, wherein the artificial potential field is composed of an improved target point gravitational potential field, an improved obstacle repulsive potential field, a road boundary potential field, and a velocity repulsive potential field;

[0009] Step S3: Calculate the fitness of the bat population, extract the optimal bat coordinates, and perform potential field processing according to their positions.

[0010] Step S4: Update the sound wave frequency, speed, and position of the bats.

[0011] Step S5: Judge whether the random number R 1 is greater than the pulse emission rate rti. If R 1 >rti, then enter the local search and obtain a new optimal solution according to different strategies; otherwise, fly randomly near the optimal solution.

[0012] Step S6: Judge whether the random number R 2 is less than the sound emission loudness Ati. If R 2 <Ati and the fitness function f(x new ) < f(xti), then accept this new optimal solution and update the sound emission loudness and pulse emission rate; otherwise, enter Step S7.

[0013] Step S7: Sort the fitness values and find the current best bat position.

[0014] Step S8: Judge whether the maximum number of iterations is reached. If the maximum number of iterations is reached, then enter Step S9; otherwise, return to Step S2.

[0015] Step S9: Smooth the path to generate the optimal path.

[0016] Furthermore, for the obstacle avoidance path planning method of an autonomous vehicle, in Step S1, establish a map environment and initialize parameters, which specifically include:

[0017] Establish a map, determine the starting point and target point of the autonomous vehicle, randomly set several obstacles (including static and dynamic ones), and initialize parameters such as the sound wave frequency, speed, position, maximum number of iterations, and pulse emission rate of each bat in the bat population.

[0018] Furthermore, for the obstacle avoidance path planning method of an autonomous vehicle, in Step S2, establish an improved artificial potential field model. The artificial potential field consists of an improved target point gravitational potential field, an improved obstacle repulsive potential field, a road boundary potential field, and a speed repulsive potential field, which specifically include:

[0019] In the artificial potential field method, when the vehicle is driving on the road, its potential energy is determined by calculating the potential energy of the gravitational potential field and multiple repulsive potential fields. The formula for establishing the improved artificial potential field model U all (X) is:

[0020] U all (X) = U att (X) + Urep (X)+U road (X)+U rev (X)

[0021] The traditional artificial potential field method has the problem that when the distance between the vehicle and the target point is far, the gravitational force is greater than the repulsive force, which may cause the vehicle to collide with the obstacle. Therefore, the improved gravitational potential field function is constructed as follows:

[0022]

[0023] In the formula, K att is the gravitational potential field gain coefficient, ρ(X,X goal ) is the Euclidean distance between the vehicle’s current position and the target point, and the vector direction is from the vehicle’s current position to the target point. g is the distance threshold.

[0024] The gravitational force is the negative gradient of the gravitational potential field, and the gravitational force is expressed as follows:

[0025]

[0026] The traditional artificial potential field method has the problem of unreachable targets. To solve this problem, the obstacle repulsion potential field is improved by adding a distance adjustment factor to the target point, that is, the distance between the current position of the vehicle and the target point. The improved obstacle repulsion potential field function is constructed as follows:

[0027]

[0028] In the formula, K rep is the repulsive potential field gain coefficient, ρ(X,X obs ) The Euclidean distance between the vehicle’s current position and the obstacle’s position, ρ 0 is the maximum impact distance of the obstacle, and n is the distance factor adjustment coefficient.

[0029] The repulsive force is the negative gradient of the repulsive potential field, and the two components of the repulsive force are obtained, and the expressions are as follows:

[0030]

[0031]

[0032] When an unmanned vehicle is actually driving, there are road boundaries and it should be constrained by the road boundaries to avoid collisions with the road boundaries. When approaching the left and right edges of the road, it should be subject to a larger repulsive force; when approaching the center of the road, the repulsive force should be relatively small. The road boundary potential field function is constructed as:

[0033]

[0034] In the formula, α 1 , α 2 is the road boundary repulsion factor, X a , X b is the horizontal coordinate of the center line of the left and right lanes, and L is the road width.

[0035] The velocity repulsive potential field function is constructed as:

[0036]

[0037] In the formula, K rev is the velocity repulsion factor, V co =||V c -V obs || represents the speed of the unmanned vehicle relative to the obstacle, and θ is the angle between the relative speed and the position vector.

[0038] Furthermore, the obstacle avoidance path planning method for an unmanned vehicle, step S3, calculates the fitness of the bat population, extracts the optimal bat coordinates and processes them into potential fields according to their positions, which specifically includes:

[0039] When planning a path, an unmanned vehicle must ensure that it does not collide with obstacles and that the path length is the shortest. To meet the above requirements, the following fitness function is constructed with the shortest path length without collision as the evaluation criterion:

[0040] (1) In order to avoid collision, an obstacle avoidance constraint penalty function is set, and the expression is as follows:

[0041]

[0042] In the formula, c is the safety factor, k is the number of obstacles, (x obs ,y obs ) are the obstacle coordinates.

[0043] (2) The path is required to be the shortest. The expression is as follows:

[0044]

[0045] In the formula, (x i ,y i ) are the coordinates of the unmanned vehicle path nodes.

[0046] In summary, the overall fitness function of an unmanned vehicle obstacle avoidance path planning method is as follows:

[0047] f=a·f 1 +b·f 2

[0048] Where a is the obstacle avoidance constraint weight coefficient and b is the path length weight coefficient.

[0049] According to the fitness function, the position coordinates of the best bat can be extracted and processed into potential fields. That is, for two adjacent path points in the bat algorithm, the gravitational fields Ut i-1 and Ut i+1 are set respectively. The expression is as follows:

[0050]

[0051]

[0052] Where xti is the current position of the ith path point at time t, is the distance between the current position and the obstacle.

[0053] Furthermore, the obstacle avoidance path planning method for an unmanned vehicle, step S4, updating the sound wave frequency, speed and position of the bat, specifically includes:

[0054] According to the bat algorithm, the bat's sound wave frequency, speed and position are updated according to the following formula:

[0055] f i =f min +(f max -f min )β

[0056]

[0057]

[0058] In the formula, f i is the frequency of the sound waves emitted by bats, and its maximum value is f max , the minimum value is f min , xti and vti represent the position and velocity of bat individual i in the tth generation, respectively, β is a random number that is uniformly distributed in [0, 1], and x* represents the global optimal position of the previous generation.

[0059] Furthermore, in the method for obstacle avoidance path planning for an unmanned vehicle, step S5 is to determine the random number R 1 Is it greater than the pulse emission rate rt i? 1 >rt i, then enter local search and obtain a new optimal solution according to different strategies; otherwise, fly randomly around the optimal solution, which specifically includes:

[0060] Generate a random number R 1 , if R 1Greater than the pulse emission rate rt i (the pulse emission rate of the i-th bat in the t-th generation), two strategies of Gaussian distribution and Cauchy distribution are used to search for local new solutions and update the position x new , the formula is as follows:

[0061]

[0062] In the formula: x old is the current optimal individual position, C(-1,0) is a random number obeying the standard Cauchy distribution; N(0,1) is a random number obeying the Gaussian distribution, A t is the average loudness of the bat in the t-th generation. P is a uniform random number in [0,1]; t is the current iteration number; T is the maximum iteration number.

[0063] In the early stage of the search, using the characteristic that the Cauchy distribution increases monotonically within its range for search can be closer to the optimal solution; in the later stage of the search, using the characteristic that the Gaussian distribution decreases monotonically within its range for search can be closer to the optimal solution. By choosing these two different search strategies, not only can the search speed be improved, but also the search accuracy can be improved.

[0064] If R 1 < rt i, then fly randomly near the optimal solution, that is, update the bat position according to the following formula:

[0065]

[0066] Furthermore, for the described obstacle avoidance path planning method for an unmanned vehicle, in step S6, it is judged whether the random number R 2 is less than the sound emission loudness At i. If R 2 < At i and the fitness function f(x new ) < f(xt i), then accept this new optimal solution and update the sound emission loudness and pulse emission rate; otherwise, enter step S7, which specifically includes:

[0067] During the process of the bat approaching the prey, the pulse emission rate will increase, and at the same time the sound emission loudness will decrease, so as to more accurately master the position of the prey. Update according to the following formula:

[0068]

[0069]

[0070] In the formula: both α and γ are constants, At+1i and rt+1i are respectively the sound emission loudness and pulse emission rate of the i-th bat in the (t + 1)-th generation, and r0 i is the initial pulse emission rate.

[0071] Furthermore, in the obstacle avoidance path planning method for an unmanned vehicle, step S8 determines whether the maximum number of iterations has been reached, and if so, proceeds to step S9, otherwise returns to step S2; step S9 smoothes the path to generate an optimal path, which specifically includes:

[0072] When the maximum number of iterations is reached, the path is smoothed using a cubic B-spline curve to generate an optimal path; otherwise, return to step S2.

[0073] The beneficial effects of the present invention are as follows:

[0074] The present invention adds an improved gravitational potential field to the target point and an improved repulsive potential field to the obstacle. At the same time, the road boundary potential field and the speed repulsive potential field are introduced to ensure that the unmanned vehicle does not collide with the road boundary and can avoid static and dynamic obstacles. When the path node is outside the maximum influence distance of the obstacle, the traditional gravitational potential field is added to the two adjacent path nodes in the bat position update formula to shorten the path length. In the update process of the local new solution of the bat algorithm, the two strategies of Gaussian distribution and Cauchy distribution are used to search and generate local new solutions, thereby improving the search speed and search accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 It is a flow chart of the present invention;

[0076] Figure 2 Modeling a map for the environment of the present invention;

[0077] Figure 3 The force analysis diagram of the obstacle repulsive potential field of the improved artificial potential field method of the present invention;

[0078] Figure 4 This is a schematic diagram of potential field processing of the present invention. DETAILED DESCRIPTION

[0079] In order to make the purpose, technical solution and advantages of the present invention clearer, the following is a further detailed description of the embodiments of the present invention in conjunction with the accompanying drawings. The specific examples described here are only used to explain the present invention and are not used to limit the present invention.

[0080] This example provides an unmanned vehicle obstacle avoidance path planning method. Figure 1 As shown, the specific steps include:

[0081] Step S1, establish a map environment and initialize parameters, which specifically includes:

[0082] Create a map, determine the starting point and target point of the driverless vehicle, and randomly set several obstacles (including static and dynamic), such as Figure 2As shown, the sound wave frequency, speed, position, maximum number of iterations, pulse emission rate and other parameters of each bat in the bat population are initialized.

[0083] Step S2, establishing an improved artificial potential field model, wherein the artificial potential field is composed of an improved target point gravitational potential field, an improved obstacle repulsive potential field, a road boundary potential field and a velocity repulsive potential field, which specifically includes:

[0084] In the artificial potential field method, when a vehicle is traveling on the road, its potential energy is determined based on the potential energy calculation of the gravitational potential field and multiple repulsive potential fields. An improved artificial potential field model U is established. all The calculation formula of (X) is:

[0085] U all (X) = U att (X)+U rep (X)+U road (X)+U rev (X)

[0086] The traditional artificial potential field method has the problem that when the distance between the vehicle and the target point is far, the gravitational force is greater than the repulsive force, which may cause the vehicle to collide with the obstacle. Therefore, the improved gravitational potential field function is constructed as follows:

[0087]

[0088] In the formula, K att is the gravitational potential field gain coefficient, ρ(X,X goal ) is the Euclidean distance between the vehicle’s current position and the target point, and the vector direction is from the vehicle’s current position to the target point. g is the distance threshold.

[0089] The gravitational force is the negative gradient of the gravitational potential field, and the gravitational force is expressed as follows:

[0090]

[0091] The traditional artificial potential field method has the problem of unreachable targets. To solve this problem, the obstacle repulsion potential field is improved, such as Figure 3 As shown, the distance adjustment factor from the target point is increased, that is, the distance between the current position of the vehicle and the target point, and the improved obstacle repulsion potential field function is constructed as follows:

[0092]

[0093] In the formula, K rep is the repulsive potential field gain coefficient, ρ(X,X obs ) The Euclidean distance between the vehicle’s current position and the obstacle’s position, ρ 0is the maximum impact distance of the obstacle, and n is the distance factor adjustment coefficient.

[0094] The repulsive force is the negative gradient of the repulsive potential field, and the two components of the repulsive force are obtained, and the expressions are as follows:

[0095]

[0096]

[0097] When an unmanned vehicle is actually driving, there are road boundaries and it should be constrained by the road boundaries to avoid collisions with the road boundaries. When approaching the left and right edges of the road, it should be subject to a larger repulsive force; when approaching the center of the road, the repulsive force should be relatively small. The road boundary potential field function is constructed as:

[0098]

[0099] In the formula, α 1 , α 2 is the road boundary repulsion factor, X a , X b is the horizontal coordinate of the center line of the left and right lanes, and L is the road width.

[0100] The velocity repulsive potential field function is constructed as:

[0101]

[0102] In the formula, K rev is the velocity repulsion factor, V co =||V c -V obs || represents the speed of the unmanned vehicle relative to the obstacle, and θ is the angle between the relative speed and the position vector.

[0103] Step S3, calculating the fitness of the bat population, extracting the optimal bat coordinates and processing them into potential fields according to their locations, which specifically includes:

[0104] When planning a path, an unmanned vehicle must ensure that it does not collide with obstacles and that the path length is the shortest. To meet the above requirements, the following fitness function is constructed with the shortest path length without collision as the evaluation criterion:

[0105] (3) In order to avoid collision, an obstacle avoidance constraint penalty function is set, and the expression is as follows:

[0106]

[0107] In the formula, c is the safety factor, k is the number of obstacles, (x obs ,y obs) are the obstacle coordinates.

[0108] (4) The path is required to be the shortest, and the expression is as follows:

[0109]

[0110] In the formula, (x i ,y i ) are the coordinates of the unmanned vehicle path nodes.

[0111] In summary, the overall fitness function of an unmanned vehicle obstacle avoidance path planning method is as follows:

[0112] f=a·f 1 +b·f 2

[0113] Where a is the obstacle avoidance constraint weight coefficient and b is the path length weight coefficient.

[0114] According to the fitness function, the position coordinates of the best bat can be extracted and processed into potential field, such as Figure 4 As shown, for two adjacent path points in the bat algorithm, the gravitational fields Ut i-1 and Ut i+1 are set respectively, and the expressions are as follows:

[0115]

[0116]

[0117] Where xti is the current position of the ith path point at time t, is the distance between the current position and the obstacle.

[0118] Step S4, updating the bat's sound wave frequency, speed and position, specifically includes:

[0119] According to the bat algorithm, the bat's sound wave frequency, speed and position are updated according to the following formula:

[0120] f i =f min +(f max -f min )β

[0121]

[0122]

[0123] In the formula, f i is the frequency of the sound waves emitted by bats, and its maximum value is f max , the minimum value is f min, \(x_{t}^{i}\) and \(v_{t}^{i}\) represent the position and velocity of bat individual \(i\) at the \(t\)-th generation respectively, \(\beta\) is a random number that is uniformly distributed in \([0, 1]\), and \(x^{*}\) represents the global best position of the previous generation.

[0124] Step S5, determine the random number \(R\) 1 whether it is greater than the pulse emission rate \(r_{t}^{i}\). If \(R\) 1 \(> r_{t}^{i}\), then enter the local search and obtain a new optimal solution according to different strategies; otherwise, fly randomly near the optimal solution, which specifically includes:

[0125] Generate a random number \(R\) 1 . If \(R\) 1 is greater than the pulse emission rate \(r_{t}^{i}\) (the pulse emission rate of the \(i\)-th bat at the \(t\)-th generation), use two strategies of Gaussian distribution and Cauchy distribution to search and generate a local new solution, and update the position \(x\) new , and the formula is as follows:

[0126]

[0127] In the formula: \(x\) old is the current optimal individual position, \(C(-1, 0)\) is a random number obeying the standard Cauchy distribution; \(N(0, 1)\) is a random number obeying the Gaussian distribution, and \(A\) t is the average loudness of the bat at the \(t\)-th generation. \(P\) is a uniform random number in \([0, 1]\); \(t\) is the current iteration number; \(T\) is the maximum iteration number.

[0128] Using the characteristic that the Cauchy distribution is monotonically increasing within its range for search in the early stage of the search can be closer to the optimal solution; using the characteristic that the Gaussian distribution is monotonically decreasing within its range for search in the later stage of the search can be closer to the optimal solution. By selecting these two different search strategies, not only can the search speed be improved, but also the search accuracy can be improved.

[0129] If \(R\) 1 \(< r_{t}^{i}\), then fly randomly near the optimal solution, that is, update the bat position according to the following formula:

[0130]

[0131] Step S6, determine whether the random number \(R\) 2 is less than the vocal loudness \(A_{t}^{i}\). If \(R\) 2 \(< A_{t}^{i}\) and the fitness function \(f(x\) new ) \(< f(x_{t}^{i})\), then accept this new optimal solution and update the vocal loudness and pulse emission rate; otherwise, enter Step S7, which specifically includes:

[0132] When bats approach their prey, the pulse emission rate will increase and the loudness will decrease, so they can more accurately grasp the location of the prey. The update formula is as follows:

[0133]

[0134]

[0135] Where: α and γ are constants, At+1i and rt+1i are the vocal loudness and pulse emission rate of the ith bat in the t+1th generation, respectively, and r0i is the initial pulse emission rate.

[0136] Step S7, sort the fitness values ​​and find the current best bat position.

[0137] Step S8, determine whether the maximum number of iterations has been reached, if so, proceed to step S9, otherwise return to step S2.

[0138] Step S9, smoothing the path and outputting the optimal path, which specifically includes:

[0139] When the maximum number of iterations is reached, the path is smoothed using a cubic B-spline curve to generate the optimal path.

Claims

1. A method for obstacle avoidance path planning for an unmanned vehicle, characterized in that: The specific steps include: Step S1, establish a map environment and initialize parameters; Step S2, establishing an improved artificial potential field model, the artificial potential field is composed of an improved target point gravitational potential field, an improved obstacle repulsive potential field, a road boundary potential field and a velocity repulsive potential field; Step S3, calculating the fitness of the bat population, extracting the optimal bat coordinates and processing them into potential fields according to their locations; Step S4, updating the bat's sound wave frequency, speed and position; Step S5, determine whether the random number R1 is greater than the pulse emission rate rt i. If R1>rt i, enter local search and obtain a new optimal solution according to different strategies; otherwise, fly randomly near the optimal solution; Step S6, determine whether the random number R2 is less than the sound emission loudness At i. If R2 < At i and the fitness function f(x new ) < f(xti), then accept this new optimal solution and update the sound emission loudness and the pulse emission rate; Otherwise, proceed to step S7; Step S7, sort the fitness values ​​to find the current best bat position; Step S8, determine whether the maximum number of iterations has been reached, if so, proceed to step S9, otherwise return to step S2; Step S9, smoothing the path to generate an optimal path.

2. The method for obstacle avoidance path planning for an unmanned vehicle according to claim 1, characterized in that: The step S1, establishing a map environment and initializing parameters, specifically includes: Create a map, determine the starting point and target point of the unmanned vehicle, randomly set several obstacles (including static and dynamic), and initialize the sound wave frequency, speed, position, maximum number of iterations, pulse emission rate and other parameters of each bat in the bat population.

3. The method for obstacle avoidance path planning for an unmanned vehicle according to claim 1, characterized in that: The step S2 is to establish an improved artificial potential field model, wherein the artificial potential field is composed of an improved target point gravitational potential field, an improved obstacle repulsive potential field, a road boundary potential field and a velocity repulsive potential field, and specifically includes: In the artificial potential field method, when a vehicle is traveling on the road, its potential energy is determined based on the potential energy calculation of the gravitational potential field and multiple repulsive potential fields. An improved artificial potential field model U is established. all The calculation formula of (X) is: U all (X)=U att (X)+U rep (X)+U road (X)+U rev (X) The traditional artificial potential field method has the problem that when the distance between the vehicle and the target point is far, the gravitational force is greater than the repulsive force, which may cause the vehicle to collide with the obstacle. Therefore, the improved gravitational potential field function is constructed as follows: In the formula, K att is the gravitational potential field gain coefficient, ρ(X,X goal ) is the Euclidean distance between the vehicle’s current position and the target point, and the vector direction is from the vehicle’s current position to the target point. g is the distance threshold; The gravitational force is the negative gradient of the gravitational potential field, and the gravitational force is expressed as follows: The traditional artificial potential field method has the problem of unreachable targets. To solve this problem, the obstacle repulsion potential field is improved by adding a distance adjustment factor to the target point, that is, the distance between the current position of the vehicle and the target point. The improved obstacle repulsion potential field function is constructed as follows: In the formula, K rep is the repulsive potential field gain coefficient, ρ(X,X obs ) The Euclidean distance between the vehicle’s current position and the obstacle’s position, ρ0 is the maximum impact distance of the obstacle, and n is the distance factor adjustment coefficient; The repulsive force is the negative gradient of the repulsive potential field, and the two components of the repulsive force are obtained, and the expressions are as follows: When an unmanned vehicle is actually driving, there are road boundaries and it should be constrained by the road boundaries to avoid collision with the road boundaries. When approaching the left and right edges of the road, it should be subject to a large repulsive force. When approaching the center of the road, the repulsive force should be relatively small. The road boundary potential field function is constructed as follows: Where α1 and α2 are the road boundary repulsion factors, X a , X b is the horizontal coordinate of the center line of the left and right lanes, and L is the road width; The velocity repulsive potential field function is constructed as: In the formula, K rev is the velocity repulsion factor, V co =||V c -V obs || represents the speed of the unmanned vehicle relative to the obstacle, and θ is the angle between the relative speed and the position vector.

4. The method for obstacle avoidance path planning for an unmanned vehicle according to claim 1, characterized in that: The step S3, calculating the fitness of the bat population, extracting the optimal bat coordinates and processing them into potential fields according to their positions, specifically includes: When planning a path, an unmanned vehicle must ensure that it does not collide with obstacles and that the path length is the shortest. To meet the above requirements, the following fitness function is constructed with the shortest path length without collision as the evaluation criterion: (1) In order to avoid collision, an obstacle avoidance constraint penalty function is set, and the expression is as follows: In the formula, c is the safety factor, k is the number of obstacles, (x obs ,y obs ) are the obstacle coordinates; (2) The path is required to be the shortest. The expression is as follows: In the formula, (x i ,y i ) are the coordinates of the unmanned vehicle path nodes; In summary, the overall fitness function of an obstacle avoidance path planning method for an autonomous vehicle is as follows: f = a·f1 + b·f2 Where a is the obstacle avoidance constraint weight coefficient and b is the path length weight coefficient; According to the fitness function, the position coordinates of the best bat can be extracted and processed by potential field, that is, for two adjacent path points in the bat algorithm, gravitational fields Ut i-1 and Ut i+1 are set respectively, and the expressions are as follows: Where xti is the current position of the i-th path point at time t, is the distance between the current position and the obstacle.

5. The method for obstacle avoidance path planning for an unmanned vehicle according to claim 1, characterized in that: In step S4, update the sound wave frequency, speed and position of its bat, which specifically includes: According to the bat algorithm, the sound wave frequency, speed and position of its bat are updated according to the following formula: f i =f min +(f max -f min )β In the formula, f i is the frequency of the sound waves emitted by bats, and its maximum value is f max , the minimum value is f min , xti and vti represent the position and velocity of bat individual i in the tth generation, respectively, β is a random number that is uniformly distributed in [0, 1], and x* represents the global optimal position of the previous generation.

6. The method for obstacle avoidance path planning for an unmanned vehicle according to claim 1, characterized in that: In step S5, judge whether the random number R1 is greater than the pulse emission rate rt i. If R1 > rt i, enter local search and obtain a new optimal solution according to different strategies; Otherwise, fly randomly near the optimal solution, which specifically includes: Generate a random number R1. If R1 is greater than the pulse emission rate rt i (the pulse emission rate of the i-th bat in the t-th generation), use the Gaussian distribution and Cauchy distribution strategies to search for a new local solution and update the position x. new , the formula is as follows: Where: x old is the current optimal individual position, C(-1,0) is a random number that obeys the standard Cauchy distribution; N(0,1) is a random number that obeys the Gaussian distribution, A t is the average loudness of the bats in the tth generation; P is a uniform random number in [0,1]; t is the current iteration number; T is the maximum iteration number; Using the property that the Cauchy distribution is monotonically increasing within its range for search in the early stage of search can be closer to the optimal solution; using the property that the Gaussian distribution is monotonically decreasing within its range for search in the later stage of search can be closer to the optimal solution; by selecting these two different search strategies, not only the search speed can be improved but also the search accuracy can be improved; If R1 < rt i, then fly randomly near the optimal solution, that is, update the bat position according to the following formula:

7. The method for obstacle avoidance path planning for an unmanned vehicle according to claim 1, characterized in that: In step S6, it is determined whether the random number R2 is less than the sound emission loudness At i. If R2 < At i and the fitness function f(x new ) < f(xt i), then this new optimal solution is accepted, and the sound emission loudness and the pulse emission rate are updated; Otherwise, enter step S7, which specifically includes: During the process of the bat approaching the prey, the pulse emission rate will increase, and at the same time the sound intensity will decrease, so that the position of the prey can be more accurately grasped, and the update formula is as follows: Where: α and γ are both constants, At+1i and rt+1i are the sound intensity and pulse emission rate of the i-th bat in the (t + 1)-th generation respectively, and r0 i is the initial pulse emission rate.

8. The method for obstacle avoidance path planning for an unmanned vehicle according to claim 1, characterized in that: In step S9, smooth the path to generate the optimal path, which specifically includes: When the maximum number of iterations is reached, use the cubic B-spline curve to smooth the path to generate the optimal path.

Images