Distributed navigation of multi-robot systems for sensing coverage

Abstract

A team of coordinating mobile robots equipped with operation specific sensors can perform different coverage tasks. If the required number of robots in the team is very large then a centralized control system becomes a complex strategy. There are also some areas where centralized communication turns into an issue. So, a team of mobile robots for coverage tasks should have the ability of decentralized or distributed decision making. This thesis investigates decentralized control of mobile robots specifically for coverage problems. A decentralized control strategy is ideally based on local information and it can offer flexibility in case there is an increment or decrement in the number of mobile robots. We perform a broad survey of the existing literature for coverage control problems. There are different approaches associated with decentralized control strategy for coverage control problems. We perform a comparative review of these approaches and use the approach based on simple local coordination rules. These locally computed nearest neighbour rules are used to develop decentralized control algorithms for coverage control problems. We investigate this extensively used nearest neighbour rule-based approach for developing coverage control algorithms. In this approach, a mobile robot gives an equal importance to every neighbour robot coming under its communication range. We develop our control approach by making some of the mobile robots playing a more influential role than other members of the team. We develop the control algorithm based on nearest neighbour rules with weighted average functions. The approach based on this control strategy becomes efficient in terms of achieving a consensus on control inputs, say heading angle, velocity, etc. The decentralized control of mobile robots can also exhibit a cyclic behaviour under some physical constraints like a quantized orientation of the mobile robot. We further investigate the cyclic behaviour appearing due to the quantized control of mobile robots under some conditions. Our nearest neighbour rule-based approach offers a biased strategy in case of cyclic behaviour appearing in the team of mobile robots. We consider a clustering technique inside the team of mobile robots. Our decentralized control strategy calculates the similarity measure among the neighbours of a mobile robot. The team of mobile robots with the similarity measure based approach becomes efficient in achieving a fast consensus like on heading angle or velocity. We perform a rigorous mathematical analysis of our developed approach. We also develop a condition based on relaxed criteria for achieving consensus on velocity or heading angle of the mobile robots. Our validation approach is based on mathematical arguments and extensive computer simulations.