A tiered planning strategy for biped navigation

Abstract

This paper presents a three-tiered planner for biped navigation over large distances through complex environments where conventional 2D planning algorithms designed for wheeled robots fail to find a solution. The lowest tier is a footstep planner which can plan sequences of footholds for navigating through obstacles or over rough terrain. The second tier is a mobile robot planner, which plans outward from the goal to the robot's initial state, building a heuristic to aid the lowest level planner in directing the footholds toward a likely path. The highest tier chooses a long-term path to follow, ignoring the details of how it will be implemented, and directs the lower levels during execution to provide footstep sequences for the robot and notification of when re-planning is necessary. Results are demonstrated with simulated environments and execution.