No abstract

No abstract

Because Autonomous Underwater Vehicles (AUVs) have limited perception and communication capabilities, designing efficient AUV flotillas is challenging. Existing solutions are often strongly related to (1) a specific kind of mission and (2) the nature of the considered AUVs. So, it is difficult to reuse these approaches when switching to another mission context. This paper proposes a generic multi-agent based layered architecture for designing and specifying AUV flotillas at a high level of abstraction, regardless of the AUVs characteristics and skills. To this end, an organizational model is used to ease and regulate interactions between heterogeneous AUVs and combined with a behavioral reactive approach for limiting communication. I. INTRODUCTION Undersea environments are particularly challenging for robotics because they are highly dynamic and constrained. In such context, most trends of research focus on Autonomous Underwater Vehicles (AUVs) because they are particularly suited for missions in which global human control can not be achieved. Considering this application domain, the environmental characteristics require taking into account three major issues. Firstly, AUV navigation is hard to achieve as the Global Position System (GPS) can not be used underwater. Some research is being conducted to allow an AUV to locate autonomously from the sea floor. For known areas, terrain referenced navigation [1] could be a solution whereas simultaneous localization and mapping (SLAM) techniques [2] are being studied to deal with unknown environments. Secondly the environment is particularly dynamic due to disturbances caused by waves, ocean currents or the inevitable moving objects and animals. Current control approaches enable effectively controlling AUV displacement. Lastly, perception and communication capabilities are very restricted underwater. Only sound waves enable long range communications. However, their uses raise numerous problems such as low data rate, multi-path, propagation delays, interferences, etc. Undersea missions typically have large areas to cover while AUV are not able to carry heavy payload because of the space and energetic limitation. Therefore, research efforts are being done on Multi-AUV approaches which rely on using several AUVs considered as a flotilla [3]. Mainly inspired by terrestrial and aerial robotics, formal approaches of centralized and decentralized control have been developed. Usually, they are focused on achieving and maintaining a particular shape for the flotilla as a whole using strategies like leader-follower [4], artificial potentials and virtual structures

No abstract