Motion planning in the belief space for compliant behaviour of a diver companion robot

Introducing robotic technology within the diving context is a very promising idea in order to improve and optimize conditions of divers during missions, both in terms of easiness and in terms of safety. In fact, a diver companion robot can support the diver by carrying bulky equipment, helping in various tasks and also correctly and promptly reacting in case of an emergency. To this aim, the Eu funded project CADDY addresses the development of a complete robotic system, made up of an Autonomous Underwater Vehicle (AUV) and an Unmanned Surface Vehicle (USV), that could effectively support divers. In this paper the particular aspect of motion planning is faced, with an emphasis on how to obtain compliance in the robot motion with respect to the diver trajectory. A promising approach is applied, namely the belief space planning strategy assuming maximum likelihood of the observations. The advantage of such a technique consists in the ability to explicitly take into account and manage measurement uncertainties which, dealing with a floating robot in an underwater environment, potentially impair task execution. The approach has been implemented and tested in simulation; preliminary results suggest the soundness of the method and its suitability to be applied in the specific underwater context.