Jong Jin Park ( 박종진 )

Applied Scientist
Amazon Lab126

My research is in the area of intelligent robot locomotion.
I want to achieve human-level decision making and motion generation in robotic systems.

Contact

Email: jongpark at amazon dot com
jongjinp at umich dot edu

Research

I work as a applied scientist at Amazon Lab126. Prior to joining Amazon Lab126, I was with Amazon Robotics, and also worked with Professor Benjamin Kuipers in the Intelligent Robotics Lab at the University of Michigan.

I work on intelligent robot locomotion. My current focus is on autonomous navigation. During my Ph.D., I worked on planning and decision-making algorithms for a passenger-carrying mobile robot, with specific application to an electric powered wheelchair navigating autonomously in indoor environments in the presence of pedestrians.

We want our robot to navigate safely and comfortably, but also confidently and efficiently in dynamic and uncertain environments while respecting the preferences of the user. This requires synthesis of nonlinear model predictive control, probabilistic decision making, robust optimization, and human-robot interaction modeling. Also, to make a robot work in real life, good debugging practice based on real data is essential.

Nonlinear control theory and C++ are my primary tools. Recently, I have started delving into deep convolutional neural net for learning non-linear dynamics. I have degrees in mechanical engineering (PhD, MSE), physics (BS), and biology (BS).

Previous Project at Michigan

Graceful Control and Navigation for an Intelligent Wheelchair

An intelligent wheelchair capable of autonomous navigation could vastly improve the quality of life and safety of hundreds of thousands of human users. At the same time, an electric powered wheelchair is a great platform for autonomous navigation research, as it is one of the smallest platform that can carry a person, have enough size and power to mount various sensors, and can operate indoors. It is also relatively easy to maintain.

I have developed the stochastic model predictive control for our robotic wheelchair, Vulcan. It identifies the current situation with available sensors, predict what is going to happen in the near future, decide and execute the best action to take at each time step. Probabilistic reasoning is essential, since real environments are inherently dynamic and uncertain due to the presence of pedestrians. Here are videos of our wheelchair robot:




The engineering contributions of this research are robust planning and control algorithms for a fully autonomous wheelchair robot that can navigate safely and gracefully in the presence of pedestrians. The long-term scientific contribution I want to make is a deeper understanding of the principles of safe, efficient, and intelligent locomotion.

Publications

Dissertation

Refereed Conferences and Workshops

Education

Ph.D., Mechanical Engineering, April 2016
University of Michigan, Ann Arbor

M.S.E., Systems and Controls, Mechanical Engineering, April 2010
University of Michigan, Ann Arbor

B.S., Physics, August 2008
Seoul National University, Seoul, Republic of Korea

B.S., Biology, August 2008
Seoul National University, Seoul, Republic of Korea

Professional Service

Program Commitee

Reviewer and Sub-reviewer

Miscellany

Citizenship: Republic of Korea (South Korea)
Erdös Number: At most 4 (Jong Jin Park -- Benjamin Kuipers (Erdos number 3))