Time: 3:00 PM, August 16th (Friday), 2024
Location: POTR 234
Coffee and bagels will be provided.
Audacious Trials toward Challenges of Adverse Weather Conditions for Autonomous Driving: Deep Learning-based Perception and Control
Yongseob Lim, Ph.D., Associate Professor, Robotics and Mechatronics Engineering, Daegu Gyeongbuk Institute of Science and Technology (DGIST), South Korea
This talk presents advancements in autonomous vehicle technologies focusing on perception, decision-making, and control. It introduces a novel dataset for evaluating object detection in various precipitation conditions using the CARLA simulator, assessing the impact on YOLO-based detectors. It also addresses night-time driving with LuminanceGAN, which enhances image brightness control and reduces artifacts, improving depth estimation with augmented KITTI datasets. Additionally, the robustness of path tracking algorithms using LSTM neural networks and model predictive control in 4-wheel independent steering systems is examined, demonstrating enhanced performance and reliability under diverse real-world conditions.
Yongseob Lim earned his Ph.D. in mechanical engineering from the University of Michigan-Ann Arbor in 2010. He has worked as a Research Engineer at Hyundai Motor Company, focusing on advanced vehicle dynamics and control technologies, and as a Principal Research Engineer at Samsung Techwin, developing stabilization algorithms for robotic arms. Currently, he is an Associate Professor in the Department of Robotics and Mechatronics Engineering at Daegu Gyeongbuk Institute of Science and Technology (DGIST). He directs the Autonomous Systems and Control Laboratory (ASCL, https://yslim73.wixsite.com/dgist-ascl). Recently, his research centers on AI-based perception, decision-making, and control for autonomous vehicles.