Human Motion Behaviour Aware Planner (HMBAP) for path planning in dynamic human environments

For a robot navigating in a human inhabited dynamic environment, the knowledge of how the robot’s movement can influence the trajectory of people around it can be very valuable. In this work we present a Human Motion Behaviour Aware Planner (HMBAP) which incorporates a Human Motion Behaviour Model (HMBM) in its planning stage to take advantage of this. HMBM is a potential field based obstacle avoidance model for people and the proposed planner uses it to give the robot a prediction of how people would react to its planned path. This information is useful for the robot to avoid imminent collisions with people in constricted spaces and the planner finds solutions in situations - called freezing robot problem - where past methods fail to find a solution. The resulting robot behaviour is also similar to how a human would move (in terms of avoidance behaviour) in a similar situation. We believe that this is a desirable feature for a robot navigating in a human inhabited environment. We have implemented HMBAP in simulation and also on the real robot in the RAMP Lab. Both simulations and experiments show the effectiveness of HMBAP.