Hybrid path planning for mobile robot using known environment model with semantic layer

Abstract

This paper presents a path planning algorithm using multiple models derived from known environment map. Each model is generated on a different abstraction level leading to a hierarchical structure; its layers include a semantic description, world topology, and metrical layer. This approach yields several benefits in control, planning and interaction aspects of the affected mobile robot. Control for the end-user is simplified as it is focused on task execution allowing for more abstract commands to be issued, which leads to a more user-friendly interface for interacting with the mobile robot. Planning process benefits from the hierarchy structure as the search space is narrowed in subsequent abstraction levels. As a consequence, less computing power is required for finding a valid route. Conducted simulations are compared with other solutions and the results demonstrate that the presented algorithm shows good performance in complicated environments.