A systematic approach to safe coordination of dynamic participants in real-time distributed systems

Abstract

Computer systems that employ autonomous robots have been demonstrated in many areas including entertainment (e.g., robot soccer), defense (e.g., reconnaissance) and homeland security (e.g., disaster rescue). To ensure successful operation, autonomous robots in these applications must coordinate their behaviors towards achieving the goals of the system while respecting safety requirements. In a safety critical system, violations of the safety requirements might endanger human safety and possibly damage crucial or expensive infrastructure, therefore, the system must be reliable despite having unreliable components (e.g., communication, sensors, or actuators). In addition, system scalability is difficult, to achieve because an increase in robot numbers results in a disproportionate increase in the required resources such as processing power, memory and communication bandwidth. Furthermore, mobile robots may be allowed to move in and out of the operational area (we term these robots ’dynamic participants’), resulting in the additional complexity of managing a varying number of robots in the environment. Indeed scalability problems are often tackled by applications resorting to using a fixed number of robots or assuming some upper bounds on robot numbers. These challenges of reliability (building a reliable system over unreliable communication and hardware), and scalability (varying robot numbers) have only been addressed jointly for specific problems.