Inverse reinforcement learning of risk-sensitive utility

The uncertain and stochastic nature of the real world poses a challenge for autonomous carsin making decisions to ensure appropriate motion, considering the safety of the passengersand other cars that may or may not be autonomous. It is crucial for these systems to learndriving patterns of other vehicles from their environment in order to predict their movementfor a better decision making. In this research, we focus on solving the highway mergingproblem, where an autonomous vehicle tries to merge onto a highway by using InverseReinforcement Learning. Human behavior is complex, and both linear and exponentialutility functions fail to capture the non-linearity associated with such decision making. Toresolve this issue, we model such behavior with a One-Switch utility function. We presentan Inverse Reinforcement Learning technique that allows an autonomous vehicle to predicthuman driving patterns to efficiently merge onto a highway by modeling risk with a one-switch utility function.