This work was featured by MIT Technology Review here.
Demonstrating the ability to train autonomous vehicles using data collected by VIRE is part of my MASc thesis. In this project, I developed a deep reinforcement learning based multi-objective autonomous braking system and trained it using data collected by VIRE, containing pedestrian road-crossing trials in which respondents walked across a road under various traffic conditions within an interactive virtual reality environment. The design of the system is formulated in a continuous action space and seeks to maximize both pedestrian safety and perception as well as passenger comfort. The policy for brake control is learned through computer simulation using two reinforcement learning algorithms i.e. Proximal Policy Optimization and Deep Deterministic Policy Gradient and the efficiency of each are compared.
Results show that the system is able to reduce the negative influence on passenger comfort by half while maintaining safe braking operation.