A Framework to Represent Joint Decisions in a Multi‐Agent Transport Simulation

Mobility simulation softwares are tools which can be used for policy evaluation or to test behavioral assumptions. While a wide range of softwares are being developped, with a wide range of approaches and assumptions, most of them focus on the representation of individual decision making. Joint decision making, on the contrary, has received much less consideration, probably due to the difficulty to represent it accurately and efficiently, and the reasonable predictions that can already be produced whithout considering it. Joint decisions may however have an important influence on mobility behaviors: household members need to coordinate to use a limited number of vehicles; groups of friends need to negociate on a place to get dinner. In particular, it is hypothesised that location choice for leisure activities is strongly influenced by the possibility to meet social contacts at destination. Thus, representing accurately such processes may allow to fill the gap observed between predictions and observations in travel distance distribution for leisure trips. This paper presents an approach to simulate such joint decision processes using the MATSim multi-agent simulation framework. It uses two main components: an approach to generate joint plans, and an approach to correlate individual-level plan choice, knowing the constraints imposed by the joint plans. The approach is designed to be usable on arbitrary social network structures. The behavior on a simple test scenario for the social network of intra-household ties is presented.