Measurement, Modeling and Prediction for Infrastructural Systems
(PI: Chandra Bhat)
The nation's surface transportation infrastructure is in urgent need of improvement from a safety and efficiency perspective. Financial constraints and rising awareness of the environmental side-effects of new highway construction is making it difficult to implement supply side strategies such as building a new highway or adding lanes to current highways. Focus has been shifting to demand side strategies (that reduce person-travel through measures such as telecommuting or reduce solo-auto travel through measures such as ridesharing incentives and congestion pricing) and to intelligent transportation system (ITS) strategies that, among other things, help the driver make more efficient decisions in a safe way. To this end, methodology must be developed to measure characteristics of infrastructural systems, model their properties and those of its users, and predict performance under various alternative scenarios.The University of Texas at Austin is involved in this multi-university collaborative effort with the National Institute of Statistical Sciences. The specific objective is to develop improved models of individuals' activity-travel decision making behavior to better predict potential individual and driver responses to transportation control strategies such as those listed above. The emphasis is on formulating realistic representations of decision-making behavior and developing efficient statistical techniques to estimate the relevant parameters of interest in these representations. Recent work under this project has led to several published papers on application of Bayesian and classical econometric methods using Gibbs sampling and related sampling techniques.
Keywords: Activity-travel behavior, infrastructure planning, statistical and mathematical methods.