ADISTRIBUTEDAPPROACHFORCOLLABORATIVETASK EXECUTIONINADYNAMICENVIRONMENT

Abstract

Autonomous mobile robots are now used in several applications, such as warehouse, factory, space, and deep-sea, that may be inaccessible for humans. The main concern is to find an effective coordination mechanism among autonomous agents to perform tasks in order to achieve high quality overall performance. Although MAS research has received substantial attention, multi-robot coordination remains a challenging problem since the overall perfor mance of the system is directly affected by the quality of coordination and control among the robots while executing cooperative tasks. Coordination in a multi-robot system can be achieved either by explicit or by implicit communication. Since explicit communication provides a better and reliable way of multi-robot coordination compared to implicit commu nication, so it is preferred in critical missions, such as search and rescue, where efficient and continuous coordination between robots is required. Acollaborative task cannot be executed by any single agent. It requires multiple agents at the task’s location. Execution of such tasks is quite challenging in a dynamic environment, as the time and location of a task arrival, required skills, and the number of robots required for its execution may not be known a priori. This necessitates the design of a distributed algorithm for collaborative task execution via runtime team/coalition formation. To form a team with a lack of global knowledge, the robots need to communicate with each other to acquire relevant information. In this thesis, a distributed approach for collaborative task execution in a dynamic environment is proposed. The algorithm is formally verified using SPIN model checker, a popular formal verification tool. We illustrate the applicability of the approach with urban search and rescue (USAR) domain and evaluate its performance with extensive experiments using ARGoS, a realistic multi-robot simulator.