子計畫五：基於雲端技術之群組機器人合作與循跡系統建置

In this research, we propose a path planning in a known indoor environment based on vector models. Because the nodes of A* algorithm are the corners of obstacle instead of nearby pixels in the vector models, an access checking operator is needed to add in the A* algorithm. This operator will examine the access between two nodes, which may be the corner of obstacle, starting point or goal. Only the adopt nodes will be considered in cost calculation while using the A* algorithm on path planning. Therefore, it ensures the route is feasible. We also propose a path smooth operator to smooth the path after the A* algorithm. This operator uses arcs to replace nodes of path, so that the path is much smoother and the mobile robot can be turned with an arc-shaped path. To avoid smoothing the path incorrectly, it also reduces the radius of gyration which is too large automatically. By using the adjusted A* algorithm and path smoother operator, the proposed method is ultimate to not only possess a good balance between shortest path and obstacle avoidance, but also provide the speed ratio by the differential wheel mobile robot model for robot motors to follow the path without situ rotation. The experiment results illustrate that the limit of motion direction problem from using A* algorithm based on grid map can be overcome, and the time spent of movement can be reduced. Due to the difference of node of composition, the propose method significantly reduces the search steps needed.