Proposal of a Graph-based Motion Planner Architecture

Abstract

Motion planning is a critical task in robots and autonomous vehicles. This task is typically complex in computation and structure. Current frameworks are difficult to extend and supervise. This article presents a new architectural proposal of motion planning software, which can be used by mobile robots and autonomous vehicles. Defining aspects have been inspired by the foundations of Cognitive Infocommunications (CogInfoCom) to construct a highly interactive system aiming intuitive human-computer collaboration. The structure is based on a graph-based approach which is suitable in a distributed setting. This formalization provides a graphical behavior description, an intuitive way for human actors to interact with the system. The behavior of each acting component is based on a hybrid-system formalism which can further serve as a basis to model human behavior.