Obin Kwon Email & Phone Number

https://obin-hero.github.io/My research interests lie in translating visual information into effective robot action policies. During my PhD, I developed novel methods for building and utilizing scene representation for robots, based on deep learning.Following items are included in my experience:(1) Extracting useful information from vision inputs using deep learning.(2) Storing such information in a form that robots can use efficiently for various tasks such as Goal-oriented navigation and Language instruction following. (SLAM, Memory Building) (3) Designing effective action policy based on the stored information using reinforcement learning and imitation learning. I also have experience in the industry as a 1) research scientist at Sequor Robotics and a 2) research intern at NAVER LABS. In both companies, I was on a specific project related to visual localization of robots.