Decentralized leader follower based formation control strategies for multiple nonholonomic mobile robots

Gamage Don, Gayan (2009) Decentralized leader follower based formation control strategies for multiple nonholonomic mobile robots. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (9MB)

Abstract

This thesis develops a hybrid decentralized formation control framework to coordinate multiple mobile robots with nonholonomic constraints. The proposed approach deploys a control theoretic bottom-up approach where, some low level behavior based controllers are coordinated by a discrete event system with supervisory control. The robots are required to navigate in an unstructured environment with a predetermined geometric formation while being adaptable to avoiding obstacles and following walls on the way. The complexity of the environment is handled by a discrete event system with supervisory control. For proper navigation, the multi robot systems are transformed in to flexible leader-follower coordinate structures, where we derive the aforementioned low level behavior based controllers. These controllers being nonlinear due to the nonholonomic nature of the robots involved, are subjected to linearization through nonlinear control techniques of static and dynamic feedback linearization. -- Trajectory tracking type formation controllers for nonholonomic mobile robots are also developed and compared against static and dynamic feedback linearized counterparts for performance. The behavior based controllers, collectively known as formation controllers, require the designated leader/leaders robot's state and velocity profiles be known to all of its followers. Hence instead of explicit communication, we use recursive Baysian estimation techniques to estimate the leader robot's state and velocity profiles through the observations taken from sensors local to the robot. We implement and simulate different recursive Baysian estimation techniques to estimate leader robot's state and compare their respective estimation accuracy. The whole conceptual system is implemented through simulation and the results are shown to verify its operation.

Actions (login required)

View Item