Why we Need a Notion of Conflict for Autonomous Traffic Agents

Abstract

Defining when autonomous traffic agents (ATAs) are in a conflict pinpoints situations where ATAs have difficulties deciding as no choice promises to achieve all their goals. This does not only relate to safety-critical decisions, for which measures must be in place to avoid or resolve a conflict. ATAs have to be aware when their choices may conflict with other values like moral or personal preferences. Taking ATAs as an example, this paper outlines how an explicit, general notion of conflict can promote the development of safe and socially acceptable autonomous agents and identifies key challenges for interdisciplinary research.