Fusion Algorithm of the Improved A* Algorithm and Segmented Bézier Curves for the Path Planning of Mobile Robots

Abstract

In terms of mobile robot path planning, the traditional A* algorithm has the following problems: a long searching time, an excessive number of redundant nodes, and too many path-turning points. As a result, the shortest path obtained from planning may not be the optimal movement route of actual robots, and it will accelerate the hardware loss of robots. To address the aforementioned problems, a fusion algorithm for path planning, combining the improved A* algorithm with segmented second-order Bézier curves, is proposed in this paper. On the one hand, the improved A* algorithm is presented to reduce unnecessary expansion nodes and shorten the search time, which was achieved from three aspects: (1) the traditional 8-neighborhood search strategy was adjusted to 5-neighborhood according to the orientation of the target point relative to the current node; (2) the dynamic weighting factor of the heuristic function was introduced into the evaluation function of the traditional A* algorithm; and (3) the key node extraction strategy was designed to reduce the redundant nodes of the optimal path. On the other hand, the optimal path planned by the improved A* algorithm was smoothed using segmented second-order Bézier curves. The simulation results show that the improved A* algorithm can effectively reduce the search time and redundant nodes and the fusion algorithm can reduce the path curvature and path length to a certain extent, improving path safety.

Research on fusing improved A* and adaptive DWA path planning algorithms

“…The experimental validation is divided into two parts. The 1st part is to use the improved neighborhood applied to the improved A* algorithm, the traditional A* algorithm, the Dijkstra algorithm, and the algorithms proposed in the literature [16], [17], [18], respectively, in a static environment, focusing on evaluating the performance in four aspects: the search time, the number of search points, the length of the paths, and the number of turning points. Part 2 is a comparison between the improved fusion algorithm, the traditional fusion algorithm [19] (traditional A* fused with traditional DWA), and the fusion algorithms proposed in the literature [20], [21] using the improved scanning areas applied to the dynamic environments, focusing on evaluating the performances in terms of four aspects: running time, path length, path security, and obstacle avoidance performance.…”

“…However, the ratio of the coefficient of the heuristic function to the actual cost function of the method is at least 1.5, which means that although the path can be found quickly, the path length may not be the shortest. On the other hand, Lai et al [17] set the weight factor of the heuristic function based on the raster map size and Manhattan distance. However, this factor mainly depends on the raster map size and fails to reflect the actual path situation accurately.…”

Research on Path Planning of UAV Swarm for Decapitation Strike

“…According to the number of control points in the path, it can be divided into first-order Bézier curves with no control points, secondorder Bézier curves with 1 control point, and third-order Bézier curves with 2 control points. The difficulty and efficiency of path planning based on Bézier curve mainly depend on the number and location of control points [29]. The second-order Bézier curve has high accuracy and short operation time, which is more suitable for path planning modeling [30].…”

Global Dynamic Path Planning of AGV Based on Fusion of Improved A* Algorithm and Dynamic Window Method

“…The global path planning of AGV is to generate the global optimal path by using the search algorithm to avoid static obstacles in the known work scene. The commonly used global path-planning algorithms include graph-based search algorithms (such as Dijkstra algorithm [ 2 ], A* algorithm [ 3 , 4 , 5 , 6 , 7 , 8 ]), sampling-based algorithms (such as rapidly exploring random tree algorithm [ 9 ]) and intelligent algorithms (such as genetic algorithm [ 10 ], ant colony algorithm [ 11 , 12 ]). Local path planning means that AGVs use sensors to collect local environment information for real-time dynamic obstacle avoidance during operation.…”

“…Wang et al [ 3 ] reduced the number of turns by introducing the length of the planned completed path and the time cost required for the planned completed path into the cost function of the A* algorithm, and then removed the nodes passing through the vertices of the obstacle and adopted the arc turning strategy to generate a safe and smooth AGV path. Lai et al [ 4 ] reduced the expansion nodes and shorten the search time by improving the neighborhood search strategy and heuristic function, deleted redundant nodes by introducing a method of preserving key nodes, and combined with a piecewise second-order Bezier curve to generate the smooth path for mobile robot driving. Zhang et al [ 5 ] introduced a deviation factor in the heuristic function to reduce the total number of expansion nodes, which is the vertical distance between the current node and the connection of starting and target point, used the bi-directional A* search strategy to speed up path search, and combined with B-spline curve.…”