A task scheduling method for agent/activity-based models

Abstract

Estimation of spatio-temporal travel demand requires accurate activity schedules as an input along with a mechanism to adapt the schedules to changing travel options. Individuals are assumed to own a duty list of activities to be accomplished within the simulated period. A partial order based on chronological and functional constraints determines the set of feasible activity execution sequences (plans). Trip and activity timing is determined by schedule prediction and adaptation. Event times in a schedule are constrained by conditions involving time-of-day (absolute time) and by duration constraints (relative time). Both types of constraints are expressed using time deviation functions (TDF). Each start and end event in a schedule induces a set of non-linear equations expressing the absolute and relative constraints. Time values are determined by solving the set of non-linear equations using a relaxation method. A discrepancy evaluation function is used both as a criterion to decide convergence of the relaxation and to compare alternative schedules for a given plan.