Robust Designs Through Risk Sensitivity:An Overview

This is an overview paper on the relationship between risk-averse designs based on exponential loss functions with or without an additional unknown(adversarial)term and some classes of stochastic games.In particular,the paper discusses the equivalences between risk-averse controller and filter designs and saddle-point solutions of some corresponding risk-neutral stochastic differential games with different information structures for the players.One of the by-products of these analyses is that risk-averse controllers and filters(or estimators)for control and signal-measurement models are robust,through stochastic dissipation inequalities,to unmodeled perturbations in controlled system dynamics as well as signal and the measurement processes.The paper also discusses equivalences between risk-sensitive stochastic zero-sum differential games and some corresponding risk-neutral three-player stochastic zero-sum differential games,as well as robustness issues in stochastic nonzero-sum differential games with finite and infinite populations of players,with the latter belonging to the domain of mean-field games.

Estimating likelihood of future crashes for crash-prone drivers

At-fault crash-prone drivers are usually future incidents or crashes. In Louisiana, considered as the high risk group for possible 34% of crashes are repeatedly committed by the at-fault crash-prone drivers who represent only 5% of the total licensed drivers in the state. This research has conducted an exploratory data analysis based on the driver faultiness and proneness. The objective of this study is to develop a crash prediction model to esti- mate the likelihood of future crashes for the at-fault drivers. The logistic regression method is used by employing eight years＇ traffic crash data （2004-2011） in Louisiana. Crash predictors such as the driver＇s crash involvement, crash and road characteristics, human factors, collision type, and environmental factors are considered in the model. The at-fault and not-at-fault status of the crashes are used as the response variable. The developed model has identified a few important variables, and is used to correctly classify at-fault crashes up to 62.40% with a specificity of 77.25%. This model can identify as many as 62.40% of the crash incidence of at-fault drivers in the upcoming year. Traffic agencies can use the model for monitoring the performance of an at-fault crash-prone drivers and making roadway improvements meant to reduce crash proneness. From the findings, it is recommended that crash-prone drivers should be targeted for special safety programs regularly through education and regulations.