Safe Autonomous Decision-Making with <i>vGOAL</i>

Abstract

Safety is one of the crucial features of autonomous systems. Safe decision-making is a critical and challenging task in developing such systems. To address this challenge, we proposed vGOAL, a GOAL-based specification language designed for ensuring safe autonomous decision-making. In this paper, we present an interpreter for vGOAL, serving as an agent-based decision-making component for autonomous systems. Our main contributions are the design and implementation of the vGOAL interpreter, which automatically and efficiently generates safe decisions in real-time, while simultaneously performing safety checking, error handling, and conflict resolution for competing requests. As the Robot Operating System is a popular framework for developing robotic application systems, we integrate the vGOAL interpreter with it via rosbridge. To demonstrate the effectiveness of vGOAL, we validated its performance using a real-world autonomous logistic system comprising three autonomous mobile robots.