Formation control of multiple mobile robots using parametric and implicit representations

Official URL: http://192.168.1.20/record=b1226357 (Table of Contents)

Coordination of autonomous robot groups is an active research area and much recent work has focused on modeling and control issues related to coordination. Robot groups can coordinate in many different ways. Some robot groups may execute coordination in which group members move in a scattered manner like the bees of a beehive or coordination of the group may require a more strict formation like the swallows. The shape formation is very important for the coordination of autonomous mobile robot groups because it increases the capability of a robot group by increasing the competence and the security of the group. The shape formation is applicable in many tasks like formation flight, flocking and schooling, transportation systems, searchand- rescue operations, competitive games, reconnaissance and surveillance. This thesis develops a flexible shape formation control method for autonomous mobile robots. There are different approaches in the literature for shape formation of mobile robots. Proposed method is different from these existing approaches by being applicable to complex formation curves as well as different number of robots and heterogeneous groups. It consists of two phases. In the first phase, shape formation is controlled by using potential fields generated from implicit polynomial representations and in the second phase, the control for keeping the desired shape is designed using elliptical Fourier descriptors. In this shape formation method, coordination between the robots is modeled using virtual linear springs between each robot and its nearest two neighbors. The success of the proposed method is shown through simulations on groups of different numbers of point-particle robots. Proposed method is then extended to non-holonomic mobile robots by using the desired positions in point particle model as references for the non-holonomic robots. The method is also implemented with real non-holonomic robots with a bird-eye-view camera.