A hierarchical topological planner for multi-storey building navigation

Abstract

In this paper, we present a hierarchical path planner for autonomous navigation of a tracked robot in buildings that have multiple storeys. The planner utilizes a multiple layered topological map that integrates both topological and metrical information to represent the multi-storey environment, which allows the robot to navigate between different storeys of the building efficiently while avoiding dynamic obstacles on the way. The high semantic levels of the map represent semantic places or areas while the low level metric graph is created on top of an occupancy grid, where the topological nodes serve as waypoints that assist the robot in the path planning. Based on the graph, the mission planner plans tasks hierarchically and each of the executing tasks is implemented as an active state in a Finite State Machine (FSM) to control the robot to complete the tasks. Simulations and realistic experiments have demonstrated the effectiveness of the proposed planner for multi-storey building navigation.