Energy Efficient Dynamic Window Approach for Local Path Planning in Mobile Service Robotics*

An energy efficient local path planner for navigation of omnidirectional battery-powered mobile robots in dynamic environments is presented in this paper. The proposed algorithm extends the Dynamic Window Approach (DWA) to incorporate a cost function based on energy consumption. The estimated energy consumption during planning is predicted using a linear regression model that is learned on the fly. Empirical results are presented on a mobile robot platform that show a 9.79% decrease in energy consumption in comparison to the DWA approach.