A Norm-Based Probabilistic Decision-Making Model for Autonomic Traffic Networks

We propose a norm-based agent-oriented model of decision-making of semi-autonomous vehicles in urban traffic scenarios. Computational norms are used to represent the driving rules and conventions that influence the distributed decisionmaking process of the vehicles. As norms restrict the admissible behaviour of the agents, we propose to represent them as constraints, and we express the agents’ individual and group decision-making in terms of distributed constraint optimization problems. The uncertain nature of the driving environment is reflected in our model through probabilistic constraints – collective norm compliance is considered as a stochastic distributed constraint optimization problem. In this paper, we introduce the basic conceptual and algorithmic ingredients of our model, including the norms provisioning and enforcement mechanisms (where electronic institutions are used), the norm semantics, as well as methods of the agents’ cooperative decision-making. For motivation and illustration of our approach, we study a cooperative multi-lane highway driving scenario; we propose a formal model, and illustrate our approach by a small example.