This paper presents a methodology to generate a group behavior for Unmanned Underwater Crawling Vehicles (UUCV) having track-style wheels. The methodology with Leader-Follower formation is developed to maximize the search efficiency of targets on three-dimensional mission field by using Fuzzy Logic. The motion of UUCV is constrained by environmental uncertainties which are represented by unknown friction coefficients between the tracks and ground surface, and an unknown topography of the operating field. The commands of all UUCV of the group in centralized mode are assigned by the central command control board or a designated group leader. In the decentralized mode, each vehicle of the group can be a leader and follower except a designated group leader and the two outmost vehicles. As a follower in this mode, each vehicle generates its own commands autonomously to maintain a constraint mutual distance and a velocity between its leader. The controller used in each UUCV, a combination of conventional Proportional-Integral-Derivative (PID) and Adaptive Neural Network Fuzzy Logic (ANNFL), is designed by using multi-objective optimization techniques from which the fixed control gains of PID controller and the adjustable control gains of ANNFL are obtained. In the simulation to implement the formation control of group vehicles, two three-dimensional spline surfaces are used for the topography of the mission area, and the friction coefficient distribution that is taken at random. When the desired command trajectory of a heading with a constant velocity and a constraining mutual distance are given, the behavior of a group of vehicles shows satisfactory results.


It has long been recognized that there are several tasks that can be performed more efficiently and robustly using multiple robots (Donald, Gariepy, and Rus, 2000; Khatib, Yokoi, Chang, 1996; Park, 2000; Rus, Donald, and Jennings, 1995; Desai, Ostrowski, and Kumar, 1998).

This content is only available via PDF.
You can access this article if you purchase or spend a download.