A real-time heterogeneous distributed computing environment for multi-robot system

Abstract

Proposes a distributed computing framework for the real-time management of a large number of robots in an industrial setting. The primary design objective of this framework is to allow a centralized heterogeneous multiprocessor controller to perform real-time scheduling and evaluation of Newton-Euler equations for manipulating the robot motion. The real-time scheduling of tasks in such an environment is shown to be an NP-hard problem, for which we propose a deadline-based heuristic. A detailed simulation reveals that the proposed heuristic can achieve near-optimal performance.