An Agent-based architecture with support to Ethical Decisions on a Road Traffic Scenario

Abstract— The road traffic scenarios with autonomous vehicles pose many challenges when automating the behaviour
of Autonomous Vehicles (AVs) to obey road traffic rules. Is it possible to assure that a given AV will always respect the road
traffic rules? More challenging, is it always the case that the vehicle is supposed to actually respect such rules? How about
those scenarios where a driver needs to break some rule to avoid potential harm to a road user? That is the reason we
need to further study the behaviour of autonomous driving in a road traffic environment. With this in mind, we use intelligent
agents to model the behaviour of an AV and add an additional ethical agent. These agents belong to architecture, where a
HITL (Human In The Loop) is also present. The HITL must be ready to retake the vehicle control in complex scenarios. In this
paper, we introduce these elements and also describe two simple examples. In further work, we shall implement our architecture
using a BDI agent programming language that gives support to the formal verification of desired properties.