Control of swarms of autonomous robots using Model Driven Development - A state-based approach

Abstract

Unmanned vehicular systems are becoming increasingly pervasive in military and civilian applications where the tedious repetitive and hazardous nature of the tasks make them indispensable. A natural progression is to bestow autonomy upon these vehicles. In this case, the resultant robots must be able to deal with unexpected circumstances on their own and, more importantly, in real-time. As a case study we focus on swarms of robots, we define as the capability of robots to keep close to each other in formation, without colliding with neighbors and obstacles. We start by modeling and simulating a possible swarm solution in MathWorks Matlab™ and, then, moving on to change the algorithm in such a way that a controller written as a Finite State Machine (FSM) may be derived. We then use IBM Rational Rose Real-Time™ (RoseRT) to implement such a controller in emulation following the formalism of Model-Driven Development (MDD).