Workshop AGV path planning method and system

Abstract

The application discloses a kind of workshop AGV path planning method and system, method includes: according to the workshop equipment layout information obtained to establish workshop grid map, in the grid map setting AGV's starting point and end point position, and mark the position of obstacle;AGV initial path is planned using D* algorithm, and the key point of initial path is extracted as the target point of AGV movement;From the target point, each corner node of AGV initial path is extracted, the starting point and end point of each corner are determined, and the motion parameters of AGV at each corner are calculated;Based on the motion parameters of each corner, the path of each corner in initial path is optimized using improved DWA algorithm, to form final path.The application can improve the running efficiency of AGV on the basis of reducing path planning time, improve transportation speed and production benefit.

Description

Technical Field

[0001] This invention relates to the field of AGV technology, specifically to a workshop AGV path planning method and system based on the fusion of D* and improved DWA algorithms. Background Technology

[0002] With the development of intelligent manufacturing technology, AGVs (Automated Guided Vehicles) have also been continuously developed and improved, resulting in more intelligent and diversified products, such as AGVs with autonomous navigation capabilities and AGVs with visual recognition technology. The popularization and application of AGVs can not only improve the production efficiency and reduce costs of enterprises, but also inject new impetus into the development of intelligent manufacturing. However, most current research on AGV path planning focuses on the shortest path, while ignoring the actual operating conditions of AGVs such as acceleration, deceleration, and turning. As a result, the transportation time of the route is not the optimal solution, and there is still room for optimization.

[0003] AGV path planning is an important problem in intelligent manufacturing. Its goal is to achieve the optimal path scheme and a specific objective function to complete specific task requirements in the most efficient way. In the process of AGV path planning, multiple factors need to be considered, such as the shortest travel distance, the shortest time, and other objectives. In the past, the research of relevant scholars at home and abroad has mainly focused on the following three aspects: (1) Path planning methods based on traditional algorithms mainly use Dijkstra's algorithm, A* algorithm, D* algorithm, etc. to plan the path of AGV. The advantages of these algorithms are that they are simple to use and have fast calculation speed, but they have the problem of poor performance in complex environments. (2) Path planning methods based on intelligent optimization algorithms mainly use deep learning, reinforcement learning, genetic algorithms, etc. to plan the path according to the running data of AGV. These algorithms can better adapt to complex environments and changing operating states, but require a large amount of data for training. (3) In addition to improving the algorithms, many scholars have started their research from the path planning problem in specific environments and designed path planning algorithms with strong adaptability, such as AGV path planning algorithms for warehousing systems and AGV path planning algorithms for hospitals. This method can better adapt to the needs of specific scenarios, but it requires in-depth research and exploration in algorithm design.

[0004] In summary, scholars both domestically and internationally have conducted in-depth research on the AGV path planning problem from different algorithmic approaches and perspectives, achieving significant progress. However, a balance between path planning speed and dynamic adaptability remains difficult to achieve, and excessive focus on the shortest path often neglects the actual operating state of the AGV, resulting in non-minimum running times. In contrast, intelligent manufacturing workshops are characterized by high dynamism and fast production pace. The workshop logistics system, as the core linking machine processing rhythms, and the AGV, as a key piece of equipment in the logistics system, require a path planning algorithm with low computational load, fast planning speed, and high dynamic adaptability. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a workshop AGV path planning method and system that can improve AGV operating efficiency by reducing planning speed and planning smooth corner paths, thereby improving workshop operation efficiency.

[0006] To achieve the above objectives, the present invention is implemented using the following technical solution:

[0007] In a first aspect, the present invention provides a workshop AGV path planning method, the method comprising:

[0008] A workshop grid map is created based on the obtained workshop equipment layout information. The start and end points of the AGVs are set on the grid map, and the positions of obstacles are marked.

[0009] The D* algorithm is used to plan the initial path of the AGV, and the key points of the initial path are extracted as the target points for the AGV movement.

[0010] Extract each corner node of the AGV's initial path from the target point, determine the start and end points of each corner, and calculate the AGV's motion parameters at each corner;

[0011] Based on the motion parameters of each corner, an improved DWA algorithm is used to optimize the path of each corner in the initial path to form the final path.

[0012] In conjunction with the first aspect, optionally, the expression for establishing the workshop grid map is as follows:

[0013]

[0014] In the formula, the size of the grid map D is M×N; D nm This represents the state value of the grid in the m-th row and n-th column, where m = 1, 2…M; n = 1, 2…N; D nm =0 indicates that the AGV can pass through this grid, D nm =1 indicates that there is an immovable obstacle in the grid, D nm =2 indicates that there is a movable obstacle in the grid.

[0015] In conjunction with the first aspect, optionally, the step of using the D* algorithm to plan the initial path of the AGV and extracting the key points of the initial path as the target points for the AGV's movement includes the following steps:

[0016] During the initial traversal, the D* algorithm is used to maintain a cost map from the end point to the starting point in reverse order, and to store the shortest path information from each point in the space to the end point.

[0017] When the map environment changes, the cost map for certain areas is modified, and an evaluation function is used to estimate the minimum cost from the starting point to the destination to obtain the initial path for the AGV; the evaluation function expression is as follows:

[0018] f(i) = g(i) + h(i)

[0019] h(i) = |x i -x goal |+|y i -y goal |

[0020] In the formula, f(i) is the total cost of the current node i; g(i) is the cost from the starting point to the current node i; h(i) is the estimated cost from the current node i to the target point; (x i ,y i ) is the position of the current node i, (x goal ,y goal ) represents the location of the target node;

[0021] The path obtained by the D* algorithm is processed by extracting key points, resulting in a series of continuous nodes P = {p1, p2, ..., p...}. k} serves as the target point for the AGV's movement.

[0022] In conjunction with the first aspect, optionally, the calculation of the AGV's motion parameters at each turning angle includes the following steps:

[0023] Turn the corner node p j The first two nodes p j-2 and the last two nodes p j+2 As the starting and ending points of this corner;

[0024] Based on the start and end points of the corner, calculate the corner node p. i The radius of curvature ρ is calculated, and the maximum linear velocity v of the AGV at that turning angle is calculated based on the radius of curvature. max and maximum angular velocity ω max The calculation formula is as follows:

[0025]

[0026]

[0027]

[0028] In the formula, L is the arc length of the path between the starting and ending points of the turning angle, Δθ is the angle in radians, g is the acceleration due to gravity, ρ is the radius of curvature, and f max f is the maximum lateral force coefficient of the AGV tire. c The coefficient of friction is the transverse friction coefficient.

[0029] In conjunction with the first aspect, optionally, the optimization of the path at each corner using the improved DWA algorithm includes the following steps:

[0030] Based on the motion parameters of the current angle to be optimized, calculate the linear velocity and angular velocity at the starting point of the angle;

[0031] Based on the linear velocity and angular velocity at the starting point of the turning angle, the DWA algorithm is used to sample the AGV's velocity at that turning angle, and the motion trajectory under each sample velocity is predicted and simulated.

[0032] The obstacle evaluation function of the overall evaluation function in the DWA algorithm is improved. When the AGV reaches the corner of the initial path, the optimal trajectory is selected from the various motion trajectories using the improved overall evaluation function, and the optimal trajectory replaces the corresponding corner part of the initial path.

[0033] By following the steps described above, we can traverse each corner of the initial path to optimize the path at each corner.

[0034] In conjunction with the first aspect, optionally, for the corner node p j The formulas for calculating the linear velocity and angular velocity at the starting point of the mid-turn angle are as follows:

[0035]

[0036]

[0037] In the formula, and These represent the starting point p of the AGV at the corner. j-2 The linear velocity and angular velocity; k1, k2, k3, k4, and k5 are the adjustment parameters of each term, Δt represents the time step; d goal and θ goal These represent the distance and angle difference between the current position of the AGV and the target position, respectively. t This represents the speed of the AGV at time step t.

[0038] In conjunction with the first aspect, optionally, the step of sampling the AGV's speed at that turning angle using the DWA algorithm includes:

[0039] The velocity sampling space V built at this turning angle d as follows:

[0040]

[0041] In the formula, represents the velocity sampling space at that turning angle; v c and w c These represent the linear velocity and angular velocity of the current AGV, respectively. vmax and awmax These represent the maximum linear acceleration and maximum angular acceleration of the AGV, respectively.

[0042] Based on velocity sampling space V d According to the preset linear velocity sampling interval E v and angular velocity E w Starting from the linear velocity and angular velocity at the beginning of the turning angle, according to the linear velocity every E... v The magnitude is taken once and the angular velocity is taken every E. w The magnitude is taken once and sampled sequentially to obtain the velocity group (v, ω).

[0043] In conjunction with the first aspect, optionally, the obstacle evaluation function of the overall evaluation function in the improved DWA algorithm includes the following steps:

[0044] Extract the estimated position nodes of the AGV and the boundary nodes of the obstacles. Calculate the distance between each estimated position node and each boundary node of the obstacles using the following formula:

[0045]

[0046] In the formula, (x s ,y s (x) represents the position coordinates of the s-th point on the obstacle boundary. p ,y p ) represents the position coordinates of the p-th estimated location node;

[0047] Iterate through the distances between each estimated position node and all nodes on the obstacle boundary, and find the shortest distance between them (Dist). min Its reciprocal is used as the obstacle evaluation function Obs c The formula is as follows:

[0048]

[0049] In conjunction with the first aspect, the optional expression for the total evaluation function G(v,ω) is as follows:

[0050] G(v,ω)=αhead(v,ω)+βObs c (v,ω)+γvel(v,ω)

[0051] In the formula, head(v,ω) and Obs c (v,ω) and vel(v,ω) are the azimuth evaluation function, obstacle evaluation function, and velocity evaluation function, respectively; α, β, and γ are the weighting coefficients of the three evaluation functions.

[0052] Secondly, the present invention provides a workshop AGV path planning system, the system comprising:

[0053] A grid map module is used to create a grid map of the workshop based on the acquired workshop equipment layout information. The grid map sets the start and end points of the AGVs and marks the positions of obstacles.

[0054] The initial path planning module is used to plan the initial path of the AGV using the D* algorithm and extract the key points of the initial path as the target points for the AGV's movement.

[0055] The corner parameter calculation module is used to extract each corner node of the AGV's initial path from the target point, determine the start and end points of each corner, and calculate the motion parameters of the AGV at each corner.

[0056] The final path optimization module is used to optimize the path of each corner in the initial path based on the motion parameters of each corner, and to form the final path.

[0057] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0058] This invention addresses the problem that traditional DWA algorithms struggle to balance path planning speed and dynamic adaptability, and that their excessive focus on shortest paths while neglecting the actual operating status of AGVs leads to non-minimum running times. Building upon the initial path planning of the D* algorithm, this invention utilizes an improved DWA algorithm to optimize the paths at each corner of the initial path. It extracts obstacle boundary nodes through velocity space search and calculates the boundary distance between the AGV and obstacles, thereby improving the obstacle evaluation function of the DWA algorithm. This reduces computational load, lowers the computational complexity of the DWA algorithm, and further reduces AGV transportation time.

[0059] Since the initial path planned by the D* algorithm has some corner points, and these corner points affect the AGV's transportation speed during movement, this invention uses an improved DWA algorithm to optimize the corner points of the initial path. Compared with the traditional method of correcting the global path, it has better dynamic obstacle avoidance effect, smoother path, shortens AGV transportation path length and time, and improves AGV transportation efficiency. Experimental results show that this invention provides an effective solution for the efficient operation of workshop AGVs, and can improve AGV operating efficiency, transportation speed and production efficiency while reducing path planning time. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the workshop AGV path planning method provided in an embodiment of the present invention;

[0061] Figure 2 Example diagram of a workshop grid map provided in an embodiment of the present invention;

[0062] Figure 3 A comparison diagram of the effects of the workshop AGV path planning method that integrates D* and improved DWA algorithms provided in the embodiments of the present invention with the D* algorithm;

[0063] Figure 4 The structural principle block diagram of the workshop AGV path planning system provided in the embodiment of the present invention. Detailed Implementation

[0064] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations thereof. In the absence of conflict, the embodiments and technical features in the embodiments can be combined with each other.

[0065] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0066] Example 1:

[0067] Reference Figure 1 As shown in the figure, this embodiment of the invention introduces a workshop AGV path planning method, which specifically includes the following steps:

[0068] S1: Establish a workshop grid map based on the obtained workshop equipment layout information, set the start and end points of the AGV on the grid map, and mark the positions of obstacles;

[0069] Specifically, this embodiment takes a smart manufacturing workshop as the research object, extracts and simplifies the layout of the workshop, and uses a grid map method to establish a two-dimensional planar spatial scene of AGV operation in the smart manufacturing workshop and performs path planning. To prevent collisions, the AGV is regarded as a point mass, and considering the actual volume of the AGV, the radius of the obstacle boundary is extended outward by the size of the AGV. Assuming the size of the grid map is M×N, the state of the entire map can be represented by the following matrix:

[0070]

[0071] In the formula, the size of the grid map D is M×N; D nm This represents the state value of the grid in the m-th row and n-th column, where m = 1, 2…M; n = 1, 2…N; D nm =0 indicates that the AGV can pass through this grid, D nm =1 indicates that there is an immovable obstacle in the grid, D nm =2 indicates that there is a movable obstacle in the grid.

[0072] Constructing a grid map of the smart manufacturing workshop, such as Figure 2 As shown, Figure 2 The intelligent manufacturing workshop is 10×10 in size. The AGVs are placed in a Cartesian coordinate system, with the horizontal axis as the X-axis and the vertical axis as the Y-axis; this is based on the actual conditions of the objects in the intelligent manufacturing workshop. Figure 2 The black grid represents static obstacles such as workshop processing equipment and robotic arms, the gray grid represents dynamic obstacles, the white grid represents all passageways, and the circular and star-shaped grids represent the starting and ending points of the AGV, respectively.

[0073] S2: Use the D* algorithm to plan the initial path of the AGV and extract the key points of the initial path as the target points for the AGV movement;

[0074] To further explain, the D* algorithm is an incremental path planning algorithm that can dynamically update the map in a real-time environment and replan the path based on the new information. Therefore, it can be applied to real-time path planning for mobile robots such as AGVs. During the initial traversal, the D* algorithm maintains a cost map in reverse order from the endpoint to the starting point, storing the shortest path information from each point in the space to the endpoint. When the map environment changes, only the cost map of a portion of the area needs to be modified, thus greatly improving efficiency during replanning. However, when there are many obstacles in the map, updating the cost map may be slow and consume a lot of computational resources. An evaluation function can be used to select the node with the minimum cost from the starting point to the current point and the node with the minimum cost from the current point to the endpoint for expansion, aiming to quickly find the optimal path from the starting point to the endpoint. The evaluation function expression is:

[0075] f(i) = g(i) + h(i)

[0076] h(i) = |x i -x goal |+|y i -y goal |

[0077] In the formula, f(i) is the total cost of the current node i; g(i) is the cost from the starting point to the current node i; h(i) is the estimated cost from the current node i to the target point, and its magnitude affects the accuracy and speed of the algorithm; (x i ,y i ) is the position of the current node i, (x goal ,y goal ) represents the location of the target node;

[0078] Furthermore, the key points of the path obtained by the D* algorithm are extracted to obtain a series of continuous nodes P = {p1, p2, ... p... k} serves as the target point for the AGV's movement.

[0079] S3: Extract each corner node of the AGV's initial path from the target point, determine the start and end points of each corner, and calculate the motion parameters of the AGV at each corner;

[0080] It should be noted that due to the complex production environment and dynamic map changes in the workshop, AGV path planning needs to meet requirements such as efficiency, reliability, and real-time performance. Considering the actual operation of AGVs, the D* algorithm results in many turns and is not smooth enough, leading to a mismatch between the path result and the actual AGV trajectory. If the AGV's running path is made to be completely consistent with the path result obtained by the D* algorithm, then the AGV needs to decelerate, stop, turn, and then accelerate again at the path turns, resulting in a non-optimal solution in terms of running time. The DWA algorithm considers the dynamic constraints of the AGV, and the planned path conforms to the actual operating state. It also has a certain degree of adaptability to dynamic changes in the environment. However, it is prone to getting trapped in local optima when the distance between the start and end points is large, and the planning speed is slow in complex environments. To address the above requirements and problems, this embodiment integrates and improves the traditional DWA algorithm and the D* algorithm.

[0081] To address the issue of excessive path turns in the D* algorithm's path planning, which does not conform to actual kinematic constraints, the DWA (Dynamic Wrapper Approach) is used to optimize the turn angles of the D* algorithm's path results. First, the D* algorithm is used to determine the path under the workshop map, resulting in a series of continuous nodes P = {p1, p2, ... p...}. k Then, extract the nodes before and after the corner of the path to obtain the set of corner nodes C = {p}. j}, where j is the index of the corner node, and finally based on each corner node p j Find the two preceding nodes p of this node. j-2 and the last two nodes p j+2 This serves as both the starting and ending point of the DWA algorithm.

[0082] In addition, the DWA algorithm also requires the direction angle and starting speed as input parameters. The direction angle is the direction of the line connecting the turning angle and the previous node, so the starting speed of the AGV also needs to be calculated. Therefore, in this embodiment, step S3 specifically includes:

[0083] S3.1: Move the corner node p j The first two nodes p j-2 and the last two nodes p j+2 As the starting and ending points of this corner;

[0084] S3.2: Based on the start and end points of the corner, calculate the corner node p. i The radius of curvature ρ is calculated, and the maximum linear velocity v of the AGV at that turning angle is calculated based on the radius of curvature. max and maximum angular velocity ωmax The calculation formula is as follows:

[0085]

[0086]

[0087]

[0088] In the formula, L is the arc length of the path between the starting and ending points of the turning angle, Δθ is the angle in radians, g is the acceleration due to gravity, ρ is the radius of curvature, and f max This is the maximum lateral force coefficient of the AGV tire, typically between 0.8 and 1.2; f c The coefficient of lateral friction is typically between 0.4 and 0.8.

[0089] S4: Based on the motion parameters of each corner, the improved DWA algorithm is used to optimize the path of each corner in the initial path to form the final path.

[0090] As an embodiment of the present invention, step S4 specifically includes:

[0091] S4.1: Based on the motion parameters of the current angle to be optimized, calculate the linear velocity and angular velocity at the starting point of the angle;

[0092] For the corner node p j The formulas for calculating the linear velocity and angular velocity at the starting point of the mid-turn angle are as follows:

[0093]

[0094]

[0095] In the formula, and These represent the starting point p of the AGV at the corner. j-2 The linear velocity and angular velocity; k1, k2, k3, k4, and k5 are the adjustment parameters of each term, Δt represents the time step; d goal and θ goal These represent the distance and angle difference between the current position of the AGV and the target position, respectively. t This represents the speed of the AGV at time step t.

[0096] S4.2: Based on the linear velocity and angular velocity at the starting point of the turning angle, the DWA algorithm is used to sample the velocity of the AGV at the turning angle, and the motion trajectory under each sample velocity is predicted and simulated.

[0097] Furthermore, the DWA algorithm considers the robot's kinematic constraints, directly searching the velocity space with obstacles, thus taking constraints into account and optimizing the robot's velocity at the next moment based on vehicle dynamics. The changes in the AGV's linear and angular velocities reflect its trajectory, and the DWA algorithm can simulate its trajectory by sampling the AGV's velocity space. Since the AGV's acceleration is limited by the motor torque, the AGV has maximum and minimum speed limitations. This embodiment constructs a velocity sampling space V at this turning angle. d as follows:

[0098]

[0099] In the formula, represents the velocity sampling space at that turning angle; v c and w c These represent the linear velocity and angular velocity of the current AGV, respectively. vmax and a wmax These represent the maximum linear acceleration and maximum angular acceleration of the AGV, respectively.

[0100] Based on velocity sampling space V d According to the preset linear velocity sampling interval E v and angular velocity E w Starting from the linear velocity and angular velocity at the beginning of the turning angle, according to the linear velocity every E... v The magnitude is taken once and the angular velocity is taken every E. w The magnitude is taken once and sampled sequentially to obtain the velocity group (v, ω).

[0101] Assuming that the AGV's displacement is small within a certain small sampling time period, and its movement trajectory is treated as a straight line, an AGV motion model is established. The mathematical expression of the AGV motion model is:

[0102]

[0103] Among them, (x t ,y t ,θ t (x) represents the current position information of the AGV. t+1 ,y t+1 ,θ t+1 ) represents the position information of the AGV at the next moment, Δt is the time step, and v x v y These are the linear velocities obtained by decomposing the current linear velocity along the X and Y axes, respectively, ω. t Let θ be the angular velocity of the AGV at the current moment. t This represents the current angle of the AGV.

[0104] S4.3: Improve the obstacle evaluation function of the overall evaluation function in the DWA algorithm. When the AGV reaches the corner of the initial path, the improved overall evaluation function is used to select the optimal trajectory from the various motion trajectories and replace the corresponding corner part of the initial path with the optimal trajectory.

[0105] Furthermore, improving the obstacle evaluation function of the overall evaluation function in the DWA algorithm includes the following steps:

[0106] S4.3.1: Extract the estimated position nodes of the AGV and the boundary nodes of the obstacles, and calculate the distance between each estimated position node and each boundary node of the obstacles using the following formula:

[0107]

[0108] In the formula, (x s ,y s (x) represents the position coordinates of the s-th point on the obstacle boundary. p ,y p ) represents the position coordinates of the p-th estimated location node;

[0109] S4.3.2: Iterate through the distances between each estimated position node and all nodes on the obstacle boundary, and find the shortest distance between them, Dist. min Its reciprocal is used as the obstacle evaluation function Obs c The formula is as follows:

[0110]

[0111] The expression for the overall evaluation function G(v,ω) is then obtained as follows:

[0112] G(v,ω)=αhead(v,ω)+βObs c (v,ω)+γvel(v,ω)

[0113] In the formula, head(v,ω) and Obs c (v,ω) and vel(v,ω) are the azimuth evaluation function, obstacle evaluation function, and velocity evaluation function, respectively; α, β, and γ are the weighting coefficients of the three evaluation functions.

[0114] Specifically, the meanings of the three evaluation functions are as follows:

[0115] head(v,ω) is the azimuth evaluation function, used to assess the error of the angle between the direction of the trajectory endpoint generated by the AGV at the current speed and the line connecting the target point. The calculation formula is as follows: in, This represents the difference between the angle between the direction of the trajectory endpoint and the line connecting the target point at the current sampling speed.

[0116] Obs c (v,ω) is an improved obstacle evaluation function used to assess the distance between the AGV's current position and the nearest obstacle. The calculation formula is:

[0117] vel(v,ω) is the speed evaluation function, which represents the current speed of the AGV. It can be directly represented by the current linear speed of the AGV.

[0118] Finally at the corner node p j At, with p j-2 Starting from p j+2 With the endpoint as the starting point and the linear velocity v and angular velocity ω as the starting velocities, the DWA algorithm is applied to solve and replace the original path results. By smoothing the path at the corners at that point, the problem of AGV spending time at corners is solved, making the robot's movement more stable.

[0119] S4.4: Follow the steps described above to traverse each corner of the initial path to optimize the path at each corner.

[0120] Specifically, from t to t max During the traversal, it is necessary to ensure that Dist in each time period is within the range of 1000 ms. c >R AGV This means the AGV will not collide with obstacles, and its speed will remain within a safe range. If a collision occurs, the AGV's movement needs to be slowed down and adjusted. Replace each corner node p. j The first two nodes p j-2 and the last two nodes p j+2 The path between points is smoothed out by smoothing the corners at those points, thus solving the problem of AGVs wasting time at corners and making the AGVs move more smoothly.

[0121] It needs further explanation that: When the traditional DWA algorithm selects the optimal (v,ω) from the velocity space, it needs to traverse every velocity value, the AGV's estimated position, and obstacle nodes. In a dynamic and complex workshop environment, the number of obstacle nodes due to people, machines, and objects results in a large search space. Generally, the size of the search space is proportional to the time complexity of path planning; the smaller the step size, the less time is required for path planning. However, too small a step size leads to low search efficiency and difficulty in finding the optimal solution. More specifically, when calculating the obstacle function, the traditional DWA algorithm calculates all possible estimated positions that the AGV can reach based on the velocity space, then traverses each position node and all obstacle nodes to obtain the estimated position node set P and the obstacle node set O, and then calculates the distance between them, resulting in a large computational load. In addition, the obstacle evaluation function of the traditional DWA algorithm requires traversing the AGV's estimated position and all obstacle node positions, resulting in a high time complexity for this process. The more complex the obstacles, the more time is consumed, which greatly affects the algorithm's computation time.

[0122] Therefore, this embodiment optimizes the traditional DWA algorithm process by starting with velocity space search and improving the way the AGV calculates the distance between itself and obstacles. It transforms the problem of solving the distance between the AGV and all nodes of the obstacle into the problem of the distance between the AGV and the obstacle boundary, thereby reducing the time complexity of the DWA algorithm during computation.

[0123] This embodiment also verifies the correctness and effectiveness of the AGV path planning based on the D* and improved DWA fusion algorithm proposed in this invention. The test simulation was performed on a computer with an Intel Core i5 (Intel(R) Core(TM) i5-6200U CPU 2.40GHz) and 8GB of RAM. The fusion algorithm was coded in Python and implemented in PyCharm 2022; details are as follows:

[0124] Before conducting simulation verification, parameter settings are required; the motion parameters of the AGV are shown in Table 1.

[0125] Table 1 AGV Motion Parameters

[0126]

[0127] The starting point is set to (0, 0), the target node is (9, 7), and dynamic obstacles are added at nodes (6, 5) and (7, 5). Path planning is then performed using both the traditional D* algorithm and the D*DWA algorithm provided in this invention, resulting in the following... Figure 3 The results of the fusion algorithm for workshop AGV path planning are shown. Figure 3 In the diagram, the black grid represents the processing equipment and robotic arms in the workshop, the gray grid represents dynamic obstacles, and the black curve represents the AGV path planning results.

[0128] Figure 3 (1) and (2) are the path results before and after using the D* algorithm. There are four turns. The AGV running time before and after the appearance of dynamic obstacles is 21.48s. Figure 3 (3) and (4) show the path results before and after using the D* and improved DWA (hereinafter referred to as D*DWA) algorithms of this invention. It can be seen that... Figure 3 (4) The solution path is smoother, and the running time of the AGV before and after the appearance of dynamic obstacles is 11.81s, which is 45% less. The reason is that the AGV needs to decelerate and turn at the corner of the path planned by the D* algorithm, which affects the transportation efficiency. As can be seen from the results, D*DWA smooths the corners solved by the D* algorithm, ensuring that it can complete the task at a near-uniform speed during transportation and reducing the time for stopping and turning.

[0129] To specifically compare the advantages of the proposed algorithm over the traditional D* algorithm, three AGV transportation tasks were randomly generated as shown in Table 2. Table 3 lists the comparison of algorithm computation time, path length, and AGV running time when solving the path using the traditional D* algorithm and the D*DWA algorithm provided by this invention.

[0130] Table 2 AGV Task List

[0131]

[0132] Table 3 Comparison of running results

[0133]

[0134] As can be seen from the comparison of the running results data in Table 3, the AGV path planning fusion algorithm based on the D*DWA algorithm of this invention optimizes the corner path in terms of path nodes compared with the traditional D* algorithm, reducing the total path length by an average of 17.17%, and the more corners the original path has, the greater the reduction in the total path length; while the AGV running time is reduced by an average of 44.79%. Therefore, the algorithm proposed in this invention enables AGVs to stop at corners less often, greatly reducing AGV running time.

[0135] In summary, the experimental results show that the workshop AGV path planning method provided by the embodiments of the present invention can ensure real-time obstacle avoidance of AGV, while shortening the transportation time of AGV and improving transportation efficiency, thus possessing certain effectiveness and advantages.

[0136] Example 2:

[0137] like Figure 4As shown, this embodiment of the invention provides a workshop AGV path planning system, which can be used to implement the method described in Embodiment 1. The system includes:

[0138] A grid map module is used to create a grid map of the workshop based on the acquired workshop equipment layout information. The grid map sets the start and end points of the AGVs and marks the positions of obstacles.

[0139] The initial path planning module is used to plan the initial path of the AGV using the D* algorithm and extract the key points of the initial path as the target points for the AGV's movement.

[0140] The corner parameter calculation module is used to extract each corner node of the AGV's initial path from the target point, determine the start and end points of each corner, and calculate the motion parameters of the AGV at each corner.

[0141] The final path optimization module is used to optimize the path of each corner in the initial path based on the motion parameters of each corner, and to form the final path.

[0142] The workshop AGV path planning system provided in this embodiment of the invention is based on the same technical concept as the workshop AGV path planning method provided in Embodiment 1, and can produce the beneficial effects described in Embodiment 1. For the contents not described in detail in this embodiment, please refer to Embodiment 1.

[0143] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0145] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction set implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0147] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A workshop AGV path planning method, characterized in that, include: A workshop grid map is created based on the obtained workshop equipment layout information. The start and end points of the AGVs are set on the grid map, and the positions of obstacles are marked. The D* algorithm is used to plan the initial path of the AGV, and the key points of the initial path are extracted as the target points for the AGV movement. Extract each corner node of the AGV's initial path from the target point, determine the start and end points of each corner, and calculate the AGV's motion parameters at each corner; Based on the motion parameters of each corner, the improved DWA algorithm is used to optimize the path of each corner in the initial path to form the final path; The calculation of the AGV's motion parameters at each turning angle includes the following steps: Turn the corner node The first two nodes and the last two nodes As the starting and ending points of this corner; Calculate the corner node based on the start and end points of the corner. radius of curvature The maximum linear velocity of the AGV at that turning angle is calculated based on the radius of curvature. and maximum angular velocity The calculation formula is as follows: ； ； ； In the formula, L is the arc length of the path between the starting and ending points of the turning angle. The angle is in radians, and g is the acceleration due to gravity. Let be the radius of curvature. This is the maximum lateral force coefficient of the AGV tire. The coefficient of lateral friction; For corner nodes The formulas for calculating the linear velocity and angular velocity at the starting point of the mid-turn angle are as follows: ； ； In the formula, and These represent the starting points of the AGV at the corners. Linear velocity and angular velocity; , , , , These are the adjustment parameters for each term. Indicates the time step; and These represent the distance from the AGV's current position to the endpoint of the current turn and the target heading angle, respectively. This represents the speed of the AGV at time step t.

2. The workshop AGV path planning method according to claim 1, characterized in that, The expression for creating the workshop grid map is as follows: ； In the formula, raster map The size is ; Indicates the first Line 1 The status values ​​of the column grid. ; This indicates that the AGV can pass through this grid. This indicates that there is an immovable obstacle in the grid. This indicates that there are movable obstacles in the grid.

3. The workshop AGV path planning method according to claim 1, characterized in that, The method of using the D* algorithm to plan the initial path of the AGV and extracting the key points of the initial path as the target points for the AGV movement includes the following steps: During the initial traversal, the D* algorithm is used to maintain a cost map from the end point to the starting point in reverse order, and to store the shortest path information from each point in the space to the end point. When the map environment changes, the cost map for certain areas is modified, and an evaluation function is used to estimate the minimum cost from the starting point to the destination to obtain the initial path for the AGV; the evaluation function expression is as follows: ； ； In the formula, The current node Total agency value; It is from the starting point to the current node. The value of the substitute; The current node The estimated cost to reach the target point; The current node Location, It is the location of the target point; The path obtained by the D* algorithm is processed by extracting key points to obtain a series of continuous nodes. , which serves as the target point for the AGV's movement.

4. The workshop AGV path planning method according to claim 1, characterized in that, The optimization of the path at each turning point using the improved DWA algorithm includes the following steps: Based on the motion parameters of the current angle to be optimized, calculate the linear velocity and angular velocity at the starting point of the angle; Based on the linear velocity and angular velocity at the starting point of the turning angle, the DWA algorithm is used to sample the AGV's velocity at that turning angle, and the motion trajectory under each sample velocity is predicted and simulated. The obstacle evaluation function of the overall evaluation function in the DWA algorithm is improved. When the AGV reaches the corner of the initial path, the optimal trajectory is selected from the various motion trajectories using the improved overall evaluation function, and the optimal trajectory replaces the corresponding corner part of the initial path. By following the steps described above, we can traverse each corner of the initial path to optimize the path at each corner.

5. The workshop AGV path planning method according to claim 4, characterized in that, The step of sampling the AGV's speed at this turning angle using the DWA algorithm includes: velocity sampling space constructed at this turning angle as follows: ； In the formula, represents the velocity sampling space at that turning angle; and These represent the current linear velocity and angular velocity of the AGV, respectively. and These represent the maximum linear acceleration and maximum angular acceleration of the AGV, respectively. Based on velocity sampling space According to the preset linear velocity sampling interval and angular velocity Starting from the linear velocity and angular velocity at the beginning of the turning angle, according to the linear velocity every other interval... The magnitude is taken once and the angular velocity is taken every other time. The magnitude is taken once and then sampled sequentially to obtain the velocity group. .

6. The workshop AGV path planning method according to claim 4, characterized in that, The obstacle evaluation function of the overall evaluation function in the improved DWA algorithm includes the following steps: Extract the estimated position nodes of the AGV and the boundary nodes of the obstacles. Calculate the distance between each estimated position node and each boundary node of the obstacles using the following formula: ； In the formula, These are the coordinates of the s-th point on the obstacle boundary. It is the first The estimated location coordinates of each node; Iterate through the distances between each estimated position node and all nodes on the obstacle boundary, and find the shortest distance between them. Its reciprocal is used as the obstacle evaluation function. The formula is as follows: 。 7. The workshop AGV path planning method according to claim 6, characterized in that, Overall evaluation function The expression is as follows: ； In the formula, These are the azimuth evaluation function, obstacle evaluation function, and velocity evaluation function, respectively. These are the weighting coefficients of the three evaluation functions.

8. A workshop AGV path planning system, characterized in that, The system employs the workshop AGV path planning method according to any one of claims 1 to 7, wherein the system comprises: A grid map module is used to create a grid map of the workshop based on the acquired workshop equipment layout information. The grid map sets the start and end points of the AGVs and marks the positions of obstacles. The initial path planning module is used to plan the initial path of the AGV using the D* algorithm and extract the key points of the initial path as the target points for the AGV's movement. The corner parameter calculation module is used to extract each corner node of the AGV's initial path from the target point, determine the start and end points of each corner, and calculate the motion parameters of the AGV at each corner. The final path optimization module is used to optimize the path of each corner in the initial path based on the motion parameters of each corner, and to form the final path.

Images