Comparison and Improvement of Local Planners on ROS for Narrow Passages

Abstract

Path planning is challenging in various extreme and complex environments due to the explosive growth of autonomous mobile robots. The widely used Dynamic Window Approach (DWA) and Time Elastic Band (TEB) local planners are integrated into Robot Operating System (ROS). A great quantity of their parameters has a significant impact on the actual performance of planning. Aiming at the scenario of robots passing through narrow passages, this work compared and optimized the typical DWA and TEB. The results of simulations showed that the improved DWA and TEB local path planners both have better performance with gentler fluctuations of velocity and smoother paths of travel in continuous narrow passages. In addition, the number of collisions was reduced to zero and the time consumption of completing navigation was decreased by an average of 15% and 7%, respectively. Moreover, the general strategies and suggestions for adjusting related parameters of DWA and TEB through narrow passages were given in this paper. Finally, the adjusted TEB local planner was applied to the customized robot in a real similar environment, and it accomplished the navigation stably with a smooth path. The results of real experiments proved the validity and reliability of our work further.