Comparison of Local Planning Algorithms for Mobile Robots

Local planning algorithms are an essential part of today's mobile robots and autonomous vehicle control. While global planning decides the route of the robot based on initial data given to the planner, local planning is a real-time motion control, based on the feedback from sensors. Its purpose is to keep the robot on an optimal track, following a plan provided by the global planner and to avoid unexpected obstacles. This makes local planning a fundamental part for safe robot navigation. The paper aims to compare three local planning algorithms: the Dynamic Window Approach, Enhanced Vector Field Histogram and Smooth Nearness-Diagram. The comparison was made on various maps in the Player/Stage system and with a real robot in SyRoTek ( System for Robotic e-learning). A large set of experiments were made at first to find best configurations for the particular planning algorithms and the robot used. After that, another set of experiments with the found parameters was conducted to gain the results needed for comparison of the algorithms. More than 20,000 simulation runs and 120 hrs of experiments with a real robot were made in total giving the results good statistical credibility.