Evaluation of an Automated Mechanism for Generating New Regulations

Humans usually use information about previous experiences to solve new problems. Following this principle, we propose an approach to enhance a multi-agent system by including an authority that generates new regulations whenever new conflicts arise. The authority uses a unsupervised version of classical Case-Based Reasoning to learn from previous similar situations and generate regulations that solve the new problem. The scenario used to illustrate and evaluate our proposal is a simulated traffic intersection where agents are traveling cars. A traffic authority observes the scenario and generates new regulations when collisions or heavy traffic are detected. At each simulation step, applicable regulations are evaluated in terms of their effectiveness and necessity in order to generate a set of regulations that, if followed, improve system performance. Empirical evaluation shows that the traffic authority succeeds in avoiding conflicting situations by automatically generating a reduced set of traffic rules.