Implementing a household joint activity-travel multi- agent simulation tool: first results

Abstract

In recent years, there has been a growing interest in the social dimension of travel, and how travel decisions are influenced not only by the global state of the transportation system, but also by joint decisions and interactions with social contacts. Such joint decisions are particularly important for a variety of behaviors: leisure activities are often performed with social contacts, and their location and timing is thus the result of a joint process; household ''maintenance'' tasks, such as grocery shopping or driving the children to school, are allocated to one single member; and of course joint (car) travel is the result of complex joint decisions. For this reason, the need to include household and friendship relationships into mobility simulation frameworks is becoming more and more obvious. This paper presents a game theoretic framework for the study of joint decision making, as well as an algorithm to search for approximate solutions of the resulting game. An implementation of this algorithm for the household case, using the MATSim software framework, is proposed and tested. The implementation of this algorithm is tested for the Zu ¨rich area, Switzerland, focusing on the possibility to perform joint trips. Those results allow to identify strengths and weaknesses of the approach, and directions for future work.

Autonomous Vehicle Fleet Sizes Required to Serve Different Levels of Demand

“…It is a well-tested scenario used in previous studies (e.g. (13) or (14)). It simulates around 2.1M agents.…”

Social interactions in transportation: analyzing groups and spatial networks

“…This kind of research can be used, for example, to forecast recreational trips within a travel behavior simulation model. Dubernet and Axhausen (2015) develop a game theoretic framework for the study of joint decision making and present a validation run where intra-household interactions are included into the MATSim activity-based, multi-agent travel microsimulation tool. Using data from the Zürich, Switzerland area and focusing on the possibility of joint trips, they find that the simulation performs reasonably well replicating the passenger mode shares for mandatory trips, but less so for discretionary trips.…”

“…Matous et al (2015) complement their quantitative analysis of a multilevel multivariate logistic model with qualitative data from classical semi-structured interviews from social science for a nuanced assessment of the impact of policies. Dubernet and Axhausen (2015) draw on co-evolutionary principles from natural computing to develop an algorithm to efficiently search approximate solutions for their framework for joint decision making, which in turn is inspired by game theory. Sharmeen et al (2015) propose a model of social network evolution inspired by life course theory and concepts from social network analysis, in a novel cross-fertilization from sociology, which is then brought to a utility-maximizing framework.…”

Modeling travel mode choice under social influence for the brazilian context

“…When someone is considered to be in your social network, another dimension is included in the urban trip phenomena. There is a shift from understanding "where people are going", "when people are going" and "what activities people are doing" towards "who they are interacting with" (Dubernet and Axhausen, 2015;Ronald et al, 2012;Hackney and Marchal, 2011;Carrasco and Miller, 2009;Axhausen, 2008).…”