| 摘要: | To guide a wheeled robot safely and smoothly in the space of obstacles, a novel path planning system including image-based planning and fuzzy path smoothing inference mechanism is proposed in this thesis. On the path planning system, various planning and smoothing techniques have been proposed to manipulate some scheduled waypoints. However, there exist some drawbacks in the announced methods, e.g., the time-consuming calculation and accumulated error etc. To tackle these problems, some obstacle reduction, path planning and smoothing methods are proposed for wheeled robot in this study. The first one is the combination of cluster reduction and Voronoi diagram to form the planned path. The fuzzy model of obstacle is used in second method. In the first stage of the proposed structures, wheeled robot captures images by a camera and the images are processed by Matlab functions. Then the reduced structure of obstacles is generated via the preprocessed image and the Voronoi diagram is utilized to plan the path in method 1. On the other hand, the membership values of obstacles are obtained firstly, then a novel path planning method is used in method 2. To guide an autonomous robot in the space of obstacles, an enhanced path planning system including image processing, cluster reduction, path planning and smoothing based on fuzzy inference is proposed in this method. To shorten the path planning time, the obstacles in the captured image are clustered into smaller groups firstly. Being a roadmap method for path generation, the fuzzy inference mechanism is employed consecutively. To smooth the planned path, a novel algorithm based on second fuzzy inference mechanism is proposed to move the rough waypoints. To verify the effectiveness of the proposed system, some simulations are carried out. From the simulations, the autonomous robot possesses good path planning and smoothing performance to reach its goal safely under various obstacle layouts. |
