This joint research program aims to develop a dynamic activity-travel assignment platform for advanced travel demand management and information system. To that effect, we combine two major fields in transportation research/operations research, i.e. activity-based modelling and dynamic traffic assignment. To accommodate the diverse choice dimensions involved in daily activity-travel patterns, the action spaces of individuals will be represented and modelled in a structural multi-state supernetwork by integrating activity programs with networks of multi-modal transport, locations of facilities\services and ICT. A derived feature is that any path through the supernetwork represents a feasible activity-travel pattern at the highest level of details. By considering the start and end points of multi-state supernetworks as virtual O-D pairs, the activity-travel assignment model will be formulated as a variational inequality problem. Thereafter, algorithms will be developed for a synthesized population to seek a state of user optimization that for each O-D pair the used path/pattern has the minimum activity-travel cost. This process will be evaluated for different scenarios concerning transport and spatial planning policies with the purpose of improving mobility efficiency and reducing the use of fuel-based private vehicles.
Individual activity-travel preferences and choice heuristics will be estimated from data gathered in large-scale surveys and travel behaviour experiments. Furthermore, a micro-simulation will be applied to assess individuals? response to day-to-day operational management. A within-day dynamic activity-travel scheduling system will be built based on the preferences from the demand side and aggregate traffic and activity participation from the supply side. The expected results will generate insights into the strategies for operational management of traffic and multi-modal public transport. This program contributes to the development of algorithms in continuous optimization for high dimensional equilibrium at a population level and in discrete optimization for dynamic scheduling at an individual level.