Offloaded Receding Horizon Planning for Environments with Variable Communication Delays

摘要

The ultimate goal for autonomous robots is quick and safe navigation in a variety of environments and under realistic conditions. Consider a robot tasked to quickly and safely navigate a cluttered environment such as a heavy forested area. One common feature of outdoor motion planning is the presence of a base station that can communicate with, plan and control the motion of this robot. This base station can offer several advantages over local motion planning, such as information about the environment that is not locally avail-able to the robot and superior computational resources. With such considerations, we propose a novel networked receding horizon planning method to navigate cluttered environments. Our proposed approach has the following capabilities: (i) it generates a sequence of waypoints optimized over a future time horizon considering knowledge about the lower level controller and the dynamics of the robot and (ii) it detects and adapts to communication delays to maintain safety. The proposed scheme is validated with simulations of a UAV flying in a cluttered forested environment under different communication losses with a base station.