A path planning method integrating an improved Harris hawk algorithm and a dynamic window method

Abstract

The present invention discloses a path planning method that combines and improves the Harris hawk algorithm and the dynamic window method, including: S1, processing the working environment; S2, initializing the Harris hawk algorithm to obtain the initial positions of the Harris hawk population; S3, calculating the fitness values of the positions of the Harris hawk individuals in the Harris hawk population, and obtaining the prey position based on the fitness values; S4, obtaining the prey escape energy through a non-linear energy factor and updating the prey position; S5, judging the current iteration number, if the current iteration number reaches the maximum iteration number, output the global path, otherwise, after adding 1 to the iteration number, return to S3; S6, obtaining the evaluation function of the dynamic window method and performing local path planning for the robot; S7, judging whether the robot reaches the target point, if it reaches, end the navigation, otherwise, return to S6. The present invention has the characteristics of real-time obstacle avoidance and optimal path.

Description

Technical Field

[0001] The present invention belongs to the technical field of path planning, and particularly relates to a path planning method that combines and improves the Harris hawk algorithm and the dynamic window method. Background Art

[0002] Aiming at the fact that traditional path planning algorithms have high mathematical requirements and cannot handle optimization problems with non-differentiable objective functions and multiple peaks. Therefore, path planning algorithms based on swarm intelligence optimization provide an effective supplement to traditional path planning algorithms in solving complex path problems.

[0003] The Harris hawk (HHO) algorithm has characteristics such as good optimization performance and fewer parameters. When the algorithm is applied to path planning, before iteration, the positions of the hawk swarm need to be initialized. Since the original model uses a random initialization method, it cannot ensure the diversity of the population, resulting in the "premature phenomenon" and search failure. In the original algorithm, the escape energy of the prey controls the switching of the hawk swarm between global search and local exploitation. The energy factor shows a linear decreasing trend. When the number of iterations increases to a certain extent, the hawk swarm only performs local search and is prone to falling into local optima.

[0004] The global path planned by using swarm intelligence optimization algorithms still cannot solve the problem of real-time obstacle avoidance of robots. Therefore, a local path planning algorithm needs to be introduced. The dynamic window method (DWA) has the characteristics of meeting the actual motion parameter requirements of mobile robots and has good local obstacle avoidance ability. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a path planning method that combines and improves the Harris hawk algorithm and the dynamic window method, which has the characteristics of taking into account both real-time obstacle avoidance and optimal path.

[0006] To achieve the above object, the present invention provides a path planning method that combines and improves the Harris hawk algorithm and the dynamic window method, including the following steps:

[0007] S1. Process the working environment;

[0008] S2. Initialize the Harris hawk algorithm, and obtain the initial positions of the Harris hawk population in combination with the processed working environment;

[0009] S3. Calculate the fitness values of the positions of Harris hawk individuals in the Harris hawk population, and obtain the prey position based on the fitness values;

[0010] S4. Obtain the escape energy of the prey through a non-linear energy factor, execute the search stage and the exploitation stage according to the escape energy of the prey, and update the prey position;

[0011] S5. Judge the current iteration number. If the current iteration number reaches the maximum iteration number, output the global path; otherwise, increment the iteration number by 1 and return to S3;

[0012] S6. Obtain the key nodes and path information in the global path. Based on the key nodes, the path information, and the trajectory evaluation function of the curvefit(v, ω) sub-function, obtain the evaluation function of the dynamic window method, and select the motion information of the optimal trajectory in the evaluation function for local path planning;

[0013] S7. Judge whether the robot reaches the target point. If it reaches, end the navigation; otherwise, return to S6 to re-perform local path planning for the robot.

[0014] Optionally, the processing of the working environment in S1 includes:

[0015] Divide the working environment into squares of the same size, and limit the search range, starting point, ending point, and obstacle information.

[0016] Optionally, the initialization of the Harris hawk algorithm in S2 includes:

[0017] Initialize the population size, iteration number, and search dimension of the Harris hawk algorithm;

[0018] Use the method of square adjacent grid proximity diffusion to obtain the initial positions of the Harris hawk population.

[0019] Optionally, the method of using the square adjacent grid proximity diffusion method to obtain the initial positions of the Harris hawk population includes:

[0020] Obtain the path starting point. Based on the path starting point, use the square proximity diffusion method to search to the map boundary or the ending point, and obtain the search dimension for initializing the positions of the Harris hawk population;

[0021] Based on the search dimension, obtain the initial positions of the Harris hawk population.

[0022] Optionally, obtaining the prey position in S3 includes:

[0023] Select the Euclidean distance as the fitness function, and calculate the fitness values of the positions of the Harris hawk individuals in the Harris hawk population through the fitness function;

[0024] Select the minimum fitness value of the positions of the Harris hawk individuals, and the minimum fitness value is the prey position.

[0025] Optionally, the method for obtaining the prey escape energy E through the non-linear energy factor in S4 is:

[0026]

[0027] Among them, E0 represents the initial energy of the prey, t represents the current iteration number of the algorithm, and T represents the maximum iteration number of the algorithm.

[0028] Optionally, the method of performing the search phase and the development phase according to the prey escape energy in S4 and updating the position of the prey includes:

[0029] When the prey escape energy |E|≥1, the search phase is executed, and the position of the prey is updated according to the states of whether the prey is discovered or not discovered during the search phase;

[0030] When the prey escape energy |E|<1, the development phase is executed, and four strategies of hard encirclement, hard siege, soft encirclement, and soft siege are respectively adopted to update the position of the prey according to the prey escape energy and the possibility of escaping within a specified range.

[0031] Optionally, after outputting the global path if the current iteration number reaches the maximum iteration number in S5, it further includes: deleting redundant nodes using collinearity and jump points to optimize the global path.

[0032] Optionally, the method of obtaining the evaluation function of the dynamic window method and selecting the motion information of the optimal trajectory in the evaluation function for local path planning in S6 includes:

[0033] Obtain the pose and speed constraint conditions of the robot;

[0034] Through the speed constraint conditions, obtain a speed constraint set and perform speed sampling to obtain a speed window:

[0035] Based on the pose and the speed window, obtain the simulated trajectory of the robot after a certain time;

[0036] Based on the key nodes, path information in the global path, and the motion and position information of the simulated trajectory after a certain time, obtain the trajectory evaluation function of the curvefit(v, ω) sub-function;

[0037] Based on the simulated trajectory after a certain time and the trajectory evaluation function of the curvefit(v, ω) sub-function, obtain the evaluation function of the dynamic window method, and select the motion information of the optimal trajectory in the evaluation function for local path planning.

[0038] Optionally, the evaluation function of the dynamic window method is:

[0039] G(v, ω) = σ[ε×heading(v, ω)+τ×dist(v, ω)+γ×vel(v, ω)+κ×curvefit(v, ω)]

[0040] Among them, heading(v, ω) represents the degree of coincidence between the azimuth angle of the end point of the simulated trajectory and the target point, dist(v, ω) represents the distance degree between the end point of the simulated trajectory and the nearest obstacle, vel(v, ω) represents the linear velocity of the simulated trajectory, curvefit(v, ω) represents the distance degree between the end position of the simulated trajectory and the global planned path, ε, τ, γ, and κ represent weights, and σ represents the normalization parameter of heading(v, ω), dist(v, ω), vel(v, ω), and curvefit(v, ω) in the evaluation function.

[0041] Technical effects of the present invention: The present invention proposes a path planning method that combines an improved Harris hawk algorithm and a dynamic window method. By introducing the Harris hawk algorithm, it provides an effective supplement for solving the multi-peak optimization problem of complex global paths. A square neighborhood adjacent diffusion method is proposed to initialize the position of the Harris hawk population, improving the initial traversability of the population under the condition of meeting population diversity. A non-linear energy factor is proposed to improve the conversion ratio of the Harris hawk in the search and development stages, enhancing the global search performance. The dynamic window method is introduced to improve the smoothness of the actual running path of the robot, and a dynamic window evaluation function combined with the global path is constructed to improve the problem of insufficient foresight of the dynamic window method. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings that form a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation of this application. In the drawings:

[0043] Figure 1 It is a schematic flowchart of a path planning method that combines an improved Harris hawk algorithm and a dynamic window method according to an embodiment of the present invention;

[0044] Figure 2 It is a schematic diagram of the Harris hawk population initialization method according to an embodiment of the present invention, where (a) is the neighborhood diffusion search method and (b) is the neighborhood classification method;

[0045] Figure 3 It is a comparison diagram of the improved prey escape energy of the Harris hawk algorithm according to an embodiment of the present invention, where (a) is the original prey escape energy E and (b) is the improved prey escape energy E;

[0046] Figure 4 It is a comparison diagram of the path planning of the improved Harris hawk algorithm and the particle swarm algorithm according to an embodiment of the present invention, where (a) is the comparison of path planning results and (b) is the comparison of convergence curves;

[0047] Figure 5This is a comparison chart of local path planning and integrated path planning in different environments according to the embodiments of the present invention. Among them, (a) is the simulation comparison of the local path planning algorithm in a simple environment, and (b) is the simulation comparison of the local path planning algorithm in a complex environment;

[0048] Figure 6 This is a schematic diagram of path planning in a dynamic obstacle environment according to the embodiments of the present invention. Among them, (a) is the planning result of the path integration IHHO-DWA algorithm, (b) is the speed curve, (c) is the angle curve, and (d) is the angular velocity curve. Detailed implementation manners

[0049] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0050] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0051] As Figure 1 shown, this embodiment provides a path planning method that combines and improves the Harris hawk algorithm and the dynamic window method, including the following steps:

[0052] 1) Grid the map and define the search range, starting point, ending point, and obstacle information;

[0053] 2) Initialize the population size M, the number of iterations T, and the search dimension N of the Harris hawk algorithm;

[0054] 3) Initialize the positions of the Harris hawk population by the method of square adjacent grid proximity diffusion. As Figure 2 (a) shows, the center represents the starting point of the path, the upper diagonal area represents the eight-neighborhood diffusion, the grid area represents the sixteen-neighborhood diffusion, the diamond area represents the twenty-four-neighborhood diffusion, the lower diagonal area represents the thirty-two-neighborhood diffusion, and so on. The search dimension of the initial position of the hawk population is searched downward in the way of square adjacent grid proximity diffusion until the map boundary or the ending point is reached.

[0055] The initialization formula is: initpos i (x, y) = N i (rand(i × 8))

[0056] where initpos i (x, y) represents the initial position of the hawk population in the i-th dimension, and N i(·) represents a list of position vectors in the i×8 neighborhood, and rand(·) represents a randomly generated index number used to index the list of position vectors.

[0057] To address the problem of insufficient population diversity, positive and negative position vector classifications are introduced, and N i (·) is divided into and As shown in Figure 2 (b), connect the starting point and the ending point and draw a perpendicular line. The eight-neighborhood nodes below the light dashed line indicate that the node is a positive node, and the eight-neighborhood nodes above the light dashed line indicate that the node is a negative node.

[0058] 4) Generate M paths to be optimized based on the population positions in step 3).

[0059] 5) Use the Euclidean distance as the fitness function, and take the minimum value of the fitness function as the prey position;

[0060]

[0061] where pos i (x, y, t) represents the set of position coordinates of the i-th individual of the Harris hawk at iteration t, f(·) represents the fitness function, and (x i，j , y i，j ) represents the position coordinates of the j-th dimension of the i-th individual of the Harris hawk in the grid map;

[0062] 6) Improve the prey escape energy E through a non-linear energy factor, and perform the search phase and the exploitation phase to update the positions;

[0063]

[0064] where E0 represents the initial energy of the prey, t represents the current iteration number of the algorithm, and T represents the maximum iteration number of the algorithm;

[0065] The comparison curve of the escape energy E of the Harris hawk algorithm when the iteration number T = 1000 is as shown in Figure 3 As can be seen from Figure 3 (b), the improved E changes slowly in the early stage of iteration, which helps the global search of the algorithm. It changes rapidly in the later stage of iteration, which helps to improve the local convergence efficiency. It can be seen that compared with the improvement in Figure 3 (a), the global search ratio has increased by 114.5%.

[0066] When |E| ≥ 1, the prey has sufficient physical strength, and the Harris hawk flock enters the search phase; the Harris hawks disperse and fly over a large area to search for prey. According to whether the prey is found or not, there are two position update states. A random number q in the range (0, 1) is generated. When q ≥ 0.5, the prey is not found, so they randomly perch within the range of the population's activities; when q < 0.5, the prey is found, and they perch based on the positions of other members and the prey. The position update formula in the search phase is as follows:

[0067]

[0068]

[0069] Among them, pos(x, y, t) represents the position vector of an individual Harris hawk at iteration t, pos(x, y, t + 1) represents the position vector of an individual Harris hawk at iteration t + 1, pos rand (x, y, t) represents the position vector of a random individual in the hawk flock, pos rabbit (x, y, t) represents the position vector of the prey, r1, r2, r3, r4 represent random numbers in the range (0, 1), ub and lb represent the upper and lower limits of the search dimension of the hawk flock, M represents the number of the Harris hawk population, pos i (x, y, t) represents the position vector of the i-th individual Harris hawk at iteration t.

[0070] When |E| < 1, the prey's physical strength drops severely, the hawk flock discovers the prey, and enters the exploitation phase; according to the prey's physical strength E and the escape probability r in the range (0, 1), four strategies are switched to capture the prey, namely siege, hard siege, soft surround, and hard surround. The position update formula is as follows:

[0071] Soft surround strategy: 0.5 ≤ |E| ≤ 1 and r ≥ 0.5

[0072] pos(x, y, t + 1) = Δpos(x, y, t) - E|Jpos rabbit (x, y, t) - pos(x, y, t)|

[0073] Δpos(x, y, t) = pos rabbit (x, y, t) - pos(x, y, t)

[0074] J = 2(1 - r5)

[0075] Among them, r5 represents a random number in the range (0, 1), Δpos(x, y, t) represents the difference between the prey and the current individual position vector at the iteration time t, pos(x, y, t + 1) represents the Harris hawk individual position vector at the iteration time t + 1, J represents the jump energy of the current prey, and it is a random number in the range (0, 2).

[0076] Hard siege strategy: |E| < 0.5 and r ≥ 0.5

[0077] pos(x, y, t + 1) = pos rabbit (x, y, t) - E|Δpos(x, y, t)|

[0078] Soft encirclement strategy: 0.5 ≤ |E| ≤ 1 and r < 0.5

[0079]

[0080] Y(x, y) = pos rabbit (x, y, t) - E|Jpos rabbit (x, y, t) - pos(x, y, t)|

[0081] Z(x, y) = Y(x, y) + S × LF(n)

[0082] Among them, f(x, y) represents the fitness function, n represents the search dimension of the hawk group, S represents an n-dimensional random variable, and LF(n) represents the levy flight function. The specific expression is as follows:

[0083]

[0084] Among them, u and v represent random numbers in the range (0, 1), and β represents a default constant with a value of 1.5.

[0085] Hard siege strategy: |E| < 0.5 and r < 0.5

[0086]

[0087] Y′(x, y) = pos rabbit (x, y, t) - E|Jpos rabbit (x, y, t) - pos m (x, y, t)|

[0088] Z′(x, y) = Y′(x, y) + S × LF(n)

[0089] 7) Calculate the fitness function values of the updated hawk group individuals (the updated hawk group individuals are represented by pos(x, y, t), specifically the (x, y) position vector at the iteration time t), and update the prey position.

[0090] 8) The iteration number \(t = t + 1\). Determine whether the current iteration number has reached the maximum iteration number \(T\). If it has reached, the iteration ends and the path with the optimal fitness value is output. Otherwise, return to step 5);

[0091] 9) Use collinearity and jump points to delete redundant nodes and optimize the global path;

[0092] Comparison of path planning between Improved Harris Hawk (IHHO) algorithm and Particle Swarm Optimization (PSO) algorithm Figure 4 As shown in the figure, a 30m×30m two-dimensional grid map is established, and the obstacle ratio is 10%. The black represents obstacles, the white represents the drivable points of the map, the dotted line represents the path planned by the improved Harris hawk algorithm, the numbers represent the key nodes of the improved Harris hawk path planning, and the solid line represents the path planned by the particle swarm optimization algorithm. The parameter settings of the two algorithms, including the initial population size, search dimension, and maximum iteration number, are all 30, 13, and 1000. Among them, the inertia factor, acceleration constant, and speed range of the particle swarm optimization algorithm are 0.9, 2, and [-2, 2] respectively. From Figure 4 (a), it can be seen that the improved Harris hawk algorithm has fewer path turning points compared to the particle swarm optimization algorithm, and the global path is smoother; from Figure 4 (b), it can be seen that the improved Harris hawk algorithm has a greater convergence amplitude than the particle swarm optimization algorithm. At the same time, the improved Harris hawk algorithm reaches the minimum iteration number after 205 iterations, and the particle swarm optimization algorithm reaches the minimum iteration number after 336 iterations. In terms of the shortest path value, the improved Harris hawk algorithm of 34.72m is better than the particle swarm optimization algorithm of 36.78m.

[0093] To prevent the contingency of the simulation results, both algorithms are run 50 times and the average values are taken. The results are shown in Table 1. It can be seen from the table that the IHHO algorithm is superior to the PSO algorithm in solving the path planning problem. The optimal path is reduced by 2.97%, the average path is reduced by 2.65%, the average iteration number is reduced by 63.49%, and the average running time of the algorithm is reduced by 41.02%. In terms of parameter settings of the two algorithms, the number of parameters of the IHHO algorithm is less than that of the PSO algorithm. Table 1 is the experimental results of the global path planning algorithm.

[0094] Table 1

[0095]

[0096] 10) Take the key nodes and path information in the global path planned by the improved Harris hawk algorithm as input and enter the local path planning stage;

[0097] 11) Initialize the robot pose and constraint conditions;

[0098] Pose update:

[0099] Among them, [x(t), y(t)] represents the position vector of the robot at time t, and θ(t), v(t), ω(t) represent the angle, velocity, and angular velocity at time t respectively, and α(t) and represent the acceleration and angular acceleration at time t respectively.

[0100] Kinematics constraint: v k = {(v, ω)|v ∈ [v min , v max , ω ∈ [ω min , ω max}

[0101] Among them, v min and v max represent the maximum and minimum values of the robot's own speed, and ω min and ω max represent the maximum and minimum values of the robot's own angular velocity.

[0102] Dynamics constraint:

[0103] Among them, v c and ω c represent the current speed and angular velocity of the robot, and represent the maximum linear deceleration and angular deceleration affected by the motor performance of the robot, and represent the maximum linear acceleration and angular acceleration affected by the motor performance of the robot.

[0104] Obstacle constraint:

[0105] Among them, dist(v, ω) represents the distance between the robot and the nearest obstacle, and this constraint ensures that the robot can stop in time before hitting the obstacle under the premise of the maximum linear deceleration and angular deceleration.

[0106] 12) According to the velocity constraint set V = v k ∩v d ∩v o , perform velocity sampling;

[0107] 13) According to the pose update in step 11) and the velocity window in step 12), simulate the movement trajectory and pose after 1.5 s;

[0108] 14) Calculate the trajectory evaluation function incorporating the curvefit(v, w) sub-function;

[0109]

[0110] Among them, \(v\) represents the speed of the simulated trajectory, \(w\) represents the angular velocity of the simulated trajectory, \((x, y)\) represents the position coordinates of the end of the simulated trajectory by the dynamic window method, \((x i , y i ) represents the coordinates of the global path guiding point at the previous moment, \((x i+1 , y i+1 ) represents the coordinates of the global path guiding point at the current moment, curvefit(v, ω) represents the distance between the end position of the simulated trajectory and the globally planned path. The larger this value is, the more the local planned path of the mobile robot can follow the globally planned path. Finally, the evaluation function of the dynamic window method is obtained as follows:

[0111] G(v, ω) = σ[ε×heading(v, ω) + τ×dist(v, ω) + γ×vel(v, ω) + κ×curvefit(v, ω)]

[0112] Among them, heading(v, ω) represents the degree of coincidence of the robot azimuth angle between the end point of the simulated trajectory and the target point, dist(v, ω) represents the distance between the end point of the simulated trajectory and the nearest obstacle, vel(v, ω) represents the linear velocity of the simulated trajectory, curvefit(v, ω) represents the distance between the end position of the simulated trajectory and the globally planned path, ε, τ, γ, and κ represent weights, and σ represents the normalization parameter of heading(v, ω), dist(v, ω), vel(v, ω), and curvefit(v, ω) in the evaluation function.

[0113] 15) Select the optimal trajectory, that is, the trajectory information with the largest evaluation function;

[0114] 16) Publish the speed, angular velocity, acceleration, angular acceleration, and position information corresponding to the optimal trajectory to the mobile robot;

[0115] 17) Determine whether the mobile robot has reached the target point. If it has reached, end the navigation; otherwise, return to step 12);

[0116] Comparison of local path planning and fusion algorithms in simple and complex environments Figure 5 As shown, in the experimental environments with obstacle ratios of 10% and 20%, a comparative experiment of the local path planning algorithm was carried out. They are the DWA algorithm without integrating global information and the IHHO-DWA algorithm that integrates global path information into the evaluation function of the dynamic window method. The parameter settings of the DWA algorithm are: robot radius size 0.3m, current direction angle Collision judgment obstacle radius Maximum linear velocity Maximum angular velocity Maximum linear acceleration Maximum angular acceleration The azimuth evaluation coefficient is 0.05, the clearance evaluation coefficient is 0.15, the speed evaluation coefficient is 0.1, and the global path evaluation coefficient is 0.5.

[0117] Finally, in an environment with an obstacle ratio of 10%, the calculation time of the Dynamic Window Approach (DWA) algorithm is 8.43 s, the actual movement time of the robot is 71.3 s, and the actual running distance is 39.03 m; the calculation time of the IHHO-DWA algorithm is 7.18 s, the actual movement time of the robot is 60.5 s, and the actual running distance is 38.45 m; Figure 5 (a) and the running results show that the DWA algorithm without global information fusion in a simple map is significantly weaker than the IHHO-DWA algorithm with global information fusion in terms of path smoothness, path value, and calculation and running time; at the same time, there are obvious turning points in the DWA algorithm without fusion near the coordinate points [13, 11] and [25, 24].

[0118] To prevent the contingency of simulation results, both algorithms are run 50 times and the average value is taken. The experimental results of the local path planning algorithm in a simple environment are shown in Table 2. The experiment shows that the IHHO-DWA algorithm is overall better than the DWA algorithm. The algorithm calculation time is improved by 8.37%, the actual movement time is improved by 16.12%, and the path value is reduced by 1.5%;

[0119] In an environment with an obstacle ratio of 20%, the calculation time of the Dynamic Window Approach (DWA) algorithm is 10.26 s, the actual movement time of the robot is 79.96 s, and the actual running distance is 48.42 m; the calculation time of the IHHO-DWA algorithm is 9.16 s, the actual movement time of the robot is 69.9 s, and the actual running distance is 45.23 m; Figure 5 (b) and the running results show that the DWA algorithm without global information fusion in a complex map is still weaker than the IHHO-DWA algorithm with global information fusion in terms of path smoothness, path value, and calculation and running time; at the same time, in terms of path selection, the IHHO-DWA algorithm with global information fusion can avoid large-scale obstacle groups, and the DWA algorithm without fusion has too many turns in the area with an abscissa of 25.

[0120] To prevent the contingency of simulation results, both algorithms are run 50 times and the average value is taken. The experimental results of the local path planning algorithm in a complex environment are shown in Table 3. The experiment shows that the IHHO-DWA algorithm is overall better than the DWA algorithm. The algorithm calculation time is improved by 11.94%, the actual movement time is improved by 15.42%, and the path value is reduced by 12.5%.

[0121] Table 2

[0122]

[0123] Table 3

[0124]

[0125] Path planning in a dynamic obstacle environment is as follows Figure 6 As shown, considering the environmental changes during the actual operation of the mobile robot, there are moving devices or personnel and adjustments to the placement of objects. Therefore, dynamic obstacles moving at a certain frequency are added to the same grid map, and static obstacles are placed on the global path for simulation to test the dynamic obstacle avoidance ability of the fusion algorithm. Figure 6 (a), the dark solid line represents the path planned by the dynamic window method, and the light dashed line represents the path planned by the improved Harris hawk algorithm. The circular dynamic obstacle coordinate points [6,5] and [16,15] appear randomly at this point and five positions above, below, left, and right of this point at a refresh frequency of 2 s; the obs marked part is the static obstacle added to the global path. When the robot starts running, since the newly added obstacle coordinate point [7,4] is close to the starting point, it chooses to detour to avoid the obstacle; when running to area A, Figure 6 (b) the speed part drops to 0, and it chooses to stop and wait; when running to area B, Figure 6 (b) there is an obvious change in the speed part, and it chooses to detour at a reduced speed; when running to area C, since the local planning tries to follow the contour of the global path as much as possible, it reduces the speed and chooses to detour near the newly added obstacle coordinate point [22,21], and in Figure 6 (c) the angle and Figure 6 (d) it can be seen from the angular velocity curve that the robot adjusts the path over a large range in an almost self-rotating manner; finally, the robot reaches the end point after running for 70.4 s, the algorithm calculation time is 7.73 s, and the path value is 38.63 m. Compared with the average value of the path fusion IHHO-DWA algorithm running 50 times in Table 2, due to the increase in the complexity of the map, the running time of the robot increases by 39.68%, the algorithm calculation time increases by 24.48%, and the path value increases by 3.45%; at the same time, compared with the path value of 36.41 m of the global path, this fusion algorithm takes into account both global optimality and dynamic obstacle avoidance in the grid map with multiple obstacle types.

[0126] The present invention aims at the realistic requirements of rapid planning, optimal path and real-time obstacle avoidance for mobile robots during the navigation process. The method of adjacent diffusion of square adjacent grids is introduced to initialize the positions of Harris hawk populations, and the forward positions are preferentially selected. A non-linear energy factor optimization conversion strategy is proposed to apply the swarm intelligence optimization algorithm to the path planning algorithm. Experiments show that compared with the PSO algorithm, the IHHO algorithm can effectively shorten the planned path and thus obtain the key nodes and node paths in the global map. The DWA algorithm introduces an evaluation function of path information, enabling the mobile robot to plan local paths in a global path priority manner. Experiments show that the IHHO-DWA path planning algorithm integrating the IHHO algorithm and the DWA algorithm realizes the actual requirements of both obstacle avoidance and optimal path.

[0127] The above are only the preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A path planning method that combines and improves the Harris hawk algorithm and the dynamic window method, characterized in that Including: S1. Process the working environment; S2. Initialize the Harris hawk algorithm, and obtain the initial position of the Harris hawk population by combining the processed working environment; The initialization of the Harris hawk algorithm in S2 includes: Initialize the population size, number of iterations, and search dimension of the Harris hawk algorithm; Use the method of square grid neighborhood diffusion to obtain the initial position of the Harris hawk population; The method of square adjacent cell proximity diffusion initializes the Harris hawk population position, including: using the center to represent the starting point of the path and performing neighborhood diffusion to search downward until the map boundary or the end point is reached. i represents the dimension; S3. Calculate the fitness value of the position of each Harris hawk individual in the Harris hawk population, and obtain the prey position based on the fitness value; S4. Obtain the prey escape energy through a non-linear energy factor, execute the search phase and exploitation phase according to the prey escape energy, and update the prey position; In S4, the method of obtaining the prey escape energy through the non-linear energy factor is as follows: : ； Among them, represents the initial energy of the prey, represents the current iteration number of the algorithm, represents the maximum iteration number of the algorithm; S5. Judge the current number of iterations. If the current number of iterations reaches the maximum number of iterations, output the global path; otherwise, add 1 to the number of iterations and return to S3; S6. Obtain the key nodes and path information in the global path, and based on the key nodes, the path information, and the trajectory evaluation function of the sub-function, obtain the evaluation function of the dynamic window method and select the motion information of the optimal trajectory in the evaluation function for local path planning; S7. Judge whether the robot reaches the target point. If it reaches, end the navigation; otherwise, return to S6 to re-plan the local path of the robot.

2. The path planning method integrating and improving the Harris hawk algorithm and the dynamic window method according to claim 1, characterized in that, The processing of the working environment in S1 includes: Divide the working environment into squares of the same size, and define the search range, starting point, ending point, and obstacle information.

3. The path planning method integrating and improving Harris hawk algorithm and dynamic window method according to claim 1, characterized in that, The method of using the square grid neighborhood diffusion method to obtain the initial position of the Harris hawk population includes: Obtain the path starting point, and based on the path starting point, use the method of square neighborhood diffusion to search to the map boundary or the ending point to obtain the search dimension for initializing the position of the Harris hawk population; Based on the search dimension, obtain the initial position of the Harris hawk population.

4. The path planning method integrating and improving Harris hawk algorithm and dynamic window method according to claim 1, characterized in that, The obtaining of the prey position in S3 includes: Select the Euclidean distance as the fitness function, and calculate the fitness value of the position of each Harris hawk individual in the Harris hawk population through the fitness function; Select the minimum fitness value of the position of the Harris hawk individual, and the minimum fitness value is the prey position.

5. The path planning method integrating and improving the Harris hawk algorithm and the dynamic window method according to claim 1, characterized in that, The method of executing the search phase and exploitation phase according to the prey escape energy and updating the prey position in S4 includes: When the escape energy of the prey is reached, the search phase is executed, and the position of the prey is updated according to the states of both discovery and non-discovery of the prey during the search phase; When the escape energy of the prey is reached, the development stage is executed. In the development stage, according to the escape energy of the prey and the possibility of escaping within a specified range, four strategies of hard encirclement, hard siege, soft encirclement, and soft siege are respectively adopted to update the position of the prey.

6. The path planning method integrating and improving Harris hawk algorithm and dynamic window method according to claim 1, characterized in that After outputting the global path when the current number of iterations reaches the maximum number of iterations in S5, it also includes: using collinearity and jump point deletion to remove redundant nodes and optimize the global path.

7. A path planning method that combines and improves the Harris hawk algorithm and the dynamic window method as described in claim 1, characterized in that, The method of obtaining the evaluation function of the dynamic window method and selecting the motion information of the optimal trajectory in the evaluation function for local path planning in S6 includes: Obtain the pose and speed constraint conditions of the robot; Through the speed constraint conditions, obtain a speed constraint set and perform speed sampling to obtain a speed window; Based on the pose and the speed window, obtain the simulated trajectory of the robot after a certain period of time; Obtain a trajectory evaluation function of a sub-function based on key nodes, path information in the global path, and the movement and position information of the simulated trajectory after a certain time of the sub-function; Based on the simulated trajectory after a certain time and the trajectory evaluation function of the sub-function, obtain the evaluation function of the dynamic window method, and select the motion information of the optimal trajectory in the evaluation function for local path planning.

8. The path planning method integrating and improving the Harris hawk algorithm and the dynamic window method according to claim 7, characterized in that, The evaluation function of the dynamic window method is: ； Among them, represents the coincidence degree of the robot azimuth angle between the end point of the simulated trajectory and the target point, represents the distance degree between the end point of the simulated trajectory and the nearest obstacle, represents the linear velocity of the simulated trajectory, represents the distance degree between the end position of the simulated trajectory and the globally planned path, 、 、 and represent weights, represents in the evaluation function 、 、 and 's normalization parameters.

Images