Autonomous hierarchy creation for computationally feasible near-optimal path planning in large environments

Abstract

Mobile robots have increasingly been used in warehouses, factories, open spaces, on roads, and urban areas to transport goods and people. Path planning in complex and large environments using classical graph-based search is computationally too intensive. One of the main challenges of mobile robotics is describing the large robot’s workspaces with a reduction of graph complexity and improvement of path planning efficiency. A representation by a hierarchical graph (H-graph) facilitates graph creation and reduces the complexity of path planning. In this paper, we present an algorithm for autonomously generating a hierarchical graph of the environment from floor plans. The hierarchical abstraction depicts the environment in levels, from the most detailed to the most abstract representation of the environment, where pre-computed partial paths at the most detailed level are graph edges at a higher level. We use the E* algorithm to find partial paths in the most detailed abstraction level, and we propose the extraction of higher levels automatically from lower levels. We verified the proposed H-graph creation on three cases: our University premises as an example of a multi-building environment, a large warehouse with a lot of shelves, and an outdoor urban environment which is represented by a city map. All cases result in five abstract levels. The result shows that the path planning algorithm searches about 50 to 800 times fewer nodes when using the H-graph as the environment representation than when using a plain graph representing the entire environment as an occupancy grid map.