Research engineer (Robotics, Deep learning) at Panasonic North America. I am especially interested in RL and control. The best part of robotics is the fact that it contains the entire categories of engineering.


I got a Master Degree in Mechanical Engineering at Columbia University. I received both my Bachelors Degree(2010) and my Master Degree(2012) in System Design Eng. from Keio University, Tokyo, Japan.

For my Masters thesis, I worked with Prof. Hod Lipson in the Creative Machines Lab and researched on developing a new type of chair (Smart Chair) to cancel out the acceleration, deceleration and centrifugal force inside the self-driving car.


Previously, I worked in the MAGLEV (Magnetic Levitated Railway) development team at Central Japan Railway Company as an Electrical/Control Engineer. I also worked in the New York office of the Founder Institute, the Silicon Valley based start-up accelerator as Operational Internship to have a insight of how to change the technology into money.

 

MY EXPERIENCE

Background & Expertise

RESEARCHER, PANASONIC NORTH AMERICA

April 2018 - Present

- Leading the development of the RL training platform (DoorGym) with UC Berkeley.
- Made the OpenAI Gym based MuJoCo+Unity AI training pipeline for a door opening task.
- Implemented the SoTA RL algos. as a baseline in the platform.
- Created the motion controller and IK & FK codes for a parallel robot in Python.
- Integrating the robot system (mechanism, sensor, AI, and controller) using ROS on Linux.
- Rapid prototyped the hardware of robotics and IoT devices using CAD, 3D printer etc.

GRADUATE RESEARCHER, CREATIVE MACHINE LAB

- Led the research of a self-stabilizing chair that cancel the external vehicle disturbances
- Designed & assembled a small-scale 3D printed & machined Smart Chair prototype
- Integrated hardware (sensor, motor) and software using ROS on Linux environment
- Developed C++ driver to sense the force with IMU and control the chair in real-time

Sep. 2016 - Mar 2018

OPERATIONS ASSOCIATE INTERN, FOUNDER INSTITUTE

- Managed the communication with 30+ start-up CEOs in the accelerator program
- Analyzed sales data of 8 major cities in the world and discovered the optimum pricing
- Attained 28% increase in enrollment in the program than 2017 by applying new pricing
- Managed the translation of the accelerator program materials from English to Japanese.

May 2017 - Mar 2018

MAGLEV CONTROL ENGINEER, CENTRAL JAPAN RAILWAY

- Conducted reliability tests of the MAGLEV autonomous driving system
- Tuned controller parameters of MAGLEV train for the test ride for the US DoT secretary
- Developed a new parameter update model and improved low speed characteristics
- Led the OJT of the member of a data analysis team and gave a monthly technical lecture

Aug 2014 – Jun 2016

PROJECT LEADER (POWER SUPPLY), CENTRAL JAPAN RAILWAY

- Led the power-supply installation project involving 9 companies, and finished on-time
- Decided device specifications and fully coordinated installation, construction and tuning
- Completed a project on time in spite of a 3-month delay by the manufacturer
- Provide technical proposals and cost reduction idea to manufacturers in meetings.

Apr 2012 – Jul 2014

 

EDUCATION

Sep 2016 - Feb 2018

MS (MECH. ENG.), COLUMBIA UNIVERSITY

- Coursework: Artificial Intelligence, Mechatronics and embedded micro, Intro to Robotics, Intro to Control Theory,  Intro to Numerical Method, Digital Control, Space Vehicle Dynamics, Lab-to-Market

Sep 2010 - Jan 2011

EXCHANGE,
LEIBNIZ UNIVERSITÄT, GERMANY

- Designed a grinding wheel that has multiple surfaces to manufacture blades for turbine engines.
- Investigated the curvature of the blade and found the optimum surface structure for the wheel.
- Collaborated with other members (Iranian and Pakistani) and efficiently delivered results.

April 2010 - Mar 2012

MS(SYSTEM DESIGN)
KEIO UNIVERISTY

Developed a component for laser printer.
-Research funded by Ricoh and Toray Dow Corning.

Aug 2006 - Mar 2010

BS (SYSTEM DESIGN) KEIO UNIVERSITY

Developed virtual tactile device.
-Devised high-accuracy experiment instrument for measuring vibration of torque.
-Proposed new procedure for material fabrication and reduced experiment time in half.

 

RESEARCH PROJECTS

DoorGym is an open source door opening task training platform to train a transferrable policy from a simulator to the real-world.

Screen Shot 2020-05-13 at 1.07.28 AM.png
 

LET’S CONNECT

 

©2020 by Yusuke Urakami. Proudly created with Wix.com