Stony Brook University Computer Science

Abstract
In this seminar, I will introduce the latest research on how robots can self-learn to drive off-road environments. For a robot to drive smoothly on unpaved roads such as off-road and walkways, it needs to identify suitable and safe areas on its own, which is challenging to achieve with traditional data-driven AI models. To overcome this, the robot must learn to recognize smooth driving areas while it’s driving. This talk will cover the latest vision-based techniques for traversable area recognition and will include insights from the speaker’s own experiences in this field.

Speaker Bio
Professor Inwook Shim is an Assistant Professor at Inha University’s Department of Smart Mobility Engineering. He holds a B.S. in Computer Science from Hanyang University (2009) and an M.S. (2011) and Ph.D. (2017) from KAIST, specializing in sensor fusion for robotic vision. Previously a Senior Researcher at Korea’s Agency for Defense Development (ADD), Professor Shim’s work focuses on autonomous navigation, traversability in complex environments, and self-supervised learning for real-time vehicle control. He received the Qualcomm Innovation Award in 2013 and contributed to Team KAIST’s DARPA Robotics Challenge win in 2015.