A problem of data
- Traditional diary surveys on travel behavior are a demanding and burdensome task for respondents, resulting in under-reporting of short trips and activities, poor data quality and falling response rates.
- The increasing availability of Big data represents a huge problem in terms of efficient data integration, data privacy and data storage.
- Big data lacks semantic interpretation, incapable of supporting the decisions of mobility and transportation management.
A problem of models
- Current model structures and assumptions are too simple: Travelers are assumed to be passive in shaping their environment in which they decide to act; action space is viewed as largely made by constraints which are present and coded a priori in the models and not by the active shaping of their context.
- Current used model outputs are insufficient: Current simulation models aggregate detailed travel behavioral data of each individual into an origin-destination matrix, from which traffic assignment algorithms are applied and travel demand is estimated. During this process, however, all potentially behaviorally rich information is lost, which is exactly the kind of information needed for the futuristic electrification scenario.
- Current models and techniques are not scalable: Today’s solutions are not suitable for representing and mining large scale data as well as simulating millions of entities in motion which require enormous volumes of CPU cycles and long executions.