In recent years, Deep Reinforcement learning has emerged as one of the most successful AI technologies by exceeding human-level performance in various games ranging from Atari to AlphaGo. However, applying reinforcement learning to real world problem is always a big challenge. This talk intends to provide a practical perspective for applying reinforcement learning to real world problems. We present a step-by-step procedure to move from gaming environment toward applying reinforcement learning for Diesel Engine Control Problem with specific challenges and their solutions. We will further discuss some of the research gaps and potential future research directions that can be beneficial for industrial applications.

Dr. Kartavya Neema serves as a Lead Engineer – Machine Learning and Controls Research at Advanced Dynamics and Controls Group at Cummins Inc. He is responsible for the development of Machine learning based Diesel Engine Controls and Calibrations that can deliver a significant performance improvement in terms of fuel economy and greenhouse gas emissions, along with significant cost reduction. Dr. Neema is a Purdue Alum and received his Ph.D and M.S. in Aeronautical and Astronautical Engineering in 2015 and 2012 respectively.