Towards ethical robots: Revisiting Braitenberg's vehicles

Abstract

The development of software and machines capable of making ethical judgements is a topic of great interest with both the research communities and the public. Debates over the possibility and practicality of such systems have only intensified with the increased use of robotics in the military arena and the ubiquity of AI in commercial products. Modern innovations, such as the driverless car, will likely make artificial ethical agents a legal necessity. As a research field, it has received relatively little attention compared to other, more traditional, AI problems. In this paper, we propose a bottom-up reactive system that provides one possible solution. We will begin by describing the motivation to this work: the development of artificial ethical agents could both mitigate some fears about the future of autonomous AI, and providing insight into human moral reasoning. We then explore the related work, including the current attempts at simulating ethics. We describe our novel approach to ethical simulation, Vessels; a Braitenberg Vehicle inspired reactive agent approach. We, then, demonstrate how Vessels can be configured to simulate both Egoism and Altruism, comparing our simulations to the normative theory.