Proscriptive Bayesian Programming Application for Collision Avoidance

Abstract

Evolve safely in an unchanged environment\nand possibly following an optimal trajectory is one big\nchallenge presented by situated robotics research field. Collision\navoidance is a basic security requirement and this\npaper proposes a solution based on a probabilistic approach\ncalled Bayesian Programming. This approach aims to deal\nwith the uncertainty, imprecision and incompleteness of the\ninformation handled. Some examples illustrate the process\nof embodying the programmer preliminary knowledge into\na Bayesian program and experimental results of these examples\nimplementation in an electrical vehicle are described\nand commented. Some videos illustrating these experiments\ncan be found at http://www-laplace.imag.fr.