directPath: An expert path planning framework to handle environment knowledge

The main purpose of this paper is to present a new path planning framework, called directPath, aimed to handle a priori environment knowledge efficiently. Our algorithm is an expert system designed to specifically exploit that knowledge, and then eventually collaborate with other frameworks in order to improve overall path planning performances. It performs a forward search with partial heuristic. It is iterative and can return a partial but safe path when terminated early. Moreover, unlike classical search algorithms, its design allows a quick detection of unfeasible problems. In this paper, we present the framework structure and illustrate its performances and properties by simulations. We will show that directPath can complete the planning task very quickly, with only minor optimality loss.