Nonlinear Stable Formation Control Using Omnidirectional Images

Abstract

This chapter has presented a multirobot formation control strategy based on nonlinear theory and omnidirectional vision. The objective is to drive a team of simple and inexpensive mobile robots to a desired formation using only visual information. Because of the limitations of the celular robots they must be led by a leader robot having the necessary processing capacity and equipped with an adequate sensor. Thus, the formation control is done using a centralized approach. Group formation is accomplished by a stable nonlinear controller designed to drive the followers into formation during navigation regardless of which sensor is used to implement the feedback. In this work, feedback was implemented using a single omnidirectional vision system because it allows the leader to localize all followers by acquiring just one image. An important advantage of our approach is that the working area is not limited since the vision system is attached to the leader and so is the reference frame, which means all measurements are taken relative to the leader. Besides that, all computations are carried out onboard the leader, discarding the use of an external computer. Through a set of image processing techniques followers' poses are reliably estimated. That includes motion segmentation, morphological filtering and color tracking, complemented by