Optimization path planning for mobile robot

Abstract

Effective and reliable path planning is essential regarding the autonomous mobile robots that work in complicated areas. This thesis proposes a faster path planning algorithm named Multi-Landmark Bidirectional A* (MLB A*) algorithm in order to achieve faster path-finding time, better navigation work in highly dynamic and partially known surroundings. Conventional A* and Bidirectional A* algorithms are useful though these algorithms have weaknesses in both computational performance and adaptability to large scale or high-density maps. The suggested MLB A* approach uses strategically positioned landmarks to exhibit the bidirectional search procedure and this will shorten the space in the search and lessen the path-finding time. The algorithm has great advantage, as it starts a forward as well as a backward search from the starting and goal positions and sets them toward intermediate landmark nodes which are nodes on the path of the goal nodes, making the search fast and efficient. In both simulated and real-world settings, experimental results show that MLB A* is more efficient in terms of faster path-finding time and success rate as compared to the standard A* and the traditional Bidirectional A* algorithms in terms of obstacle distribution complexity. This strategy is prospective in real-time robotic positioning in automation of warehouses, products delivery or service robots.