Socially Aware Local Planning: a Dynamic Window-based Approach

As the interest in mobile robots navigation on public sidewalks increases, so does the attention given to local planners capable of navigating around humans in a socially acceptable way. This paper presents a socially aware version of the Dynamic Window Approach planner. The DWA is augmented with an additional cost function, which predicts pedestrian trajectories using the Social Forces Model. Scoring of the robot control action is achieved by weighting the disturbance the robot causes to pedestrians. The approach is validated in a simulation environment with realistic pedestrian motions, showing superior performance with respect to the original DWA as well as to a distance-based scoring method.