기계공학부 구성원들의 많은 관심과 참여 부탁드립니다.

▣ 주   제: Optimal Planning and Control for Autonomous Indoor Mobility

▣ 연   사: Soo Jeon 교수

▣ 소   속: University of Waterloo

▣ 일   시: 2023. 1. 3.(Tue) 16:00

▣ 장   소: 제4공학관 D508호

▣ 초   청: 김종백 교수

▣ 초   록

Autonomous mobility plays an indispensable role in the operation of robotic agents for assisting human workers in large industrial sites such as warehouse and smart factories. The environments under which they are intended to operate are often complex and dynamic, and thus pose numerous technical challenges to ensure safety and reliability while we also pursue efficiency and performance. Optimization techniques using mathematical models of the environment and the robot have long been a go-to approach to meet (or strike a balance between) different objectives associated with safety and performance. This talk will overview some examples from recent projects on autonomous indoor mobility that lie along this paradigm. While planning and control can still (and will continue to) benefit from model-based techniques, recent advances in sensing and deep learning bring us new opportunities with data-oriented methods, especially in the area of perception. This talk will also cover an example on this viewpoint. Specific topics of this talk will include 1) the safety-aware global and local planning, 2) set point and path following control of holonomic mobile robots using nonlinear model-predictive control without terminal cost, and 3) the confidence-guided raw disparity fusion for self-supervised stereo depth estimation. For each case, key issues and attempted solutions will be presented followed by performance evaluation.