TeamK @ World Robot Summit, Japan -- 12.2019
|
After winning the DARPA Robotics Competition, our lab's vision was to expand to other areas of robotics such as manipulation. With Amazon Picking Challenge discontinued, we aimed to compete in the Stock & Disposal task at the World Robot Summit in Japan in 2020.
With 3 months prior notice, we received information about WRS Trials in 2019, and our professor composed a team to compete. This entailed creating a new mobile base for the commercial RB5 collaborative robot arm, as well as all the SW involving mission execution for the competition. As this would be my last event (near the end of my 3 year contract at KAIST), it was a bitter-sweet moment. The meticulously planned Gantt Chart kept us efficiently-focused on achieving what we could in the short preparation time. Prior to arriving in Japan, we were confident in scoring a fair amount. However, upon speaking to the judge after arrival, we learned some of the assumptions we've made (objects' orientation and clutteredness) were different and required heavy changes in code. With the best we could muster, we attempted last-minute strategies but ended-up sticking to the original one -- and placed 3rd place. |
My first international conference: IROS -- 11.2019
|
IROS definitely was a different tier conference than ROSCon or KROS (Korean robot conference). The sheer number of participants, 20 sessions running in parallel, and the depth of content presented were overwhelming. My initial crave to absorb 100% of the conference quickly changed. I instead resorted to just sitting in 1 session and getting to know familiar faces in that field. This approach led to some fortuitous friendships and networks that I hope to maintain in my future conferences as well.
Industry participation was phenomenal as well. Chinese robotic companies, such as Unitree, developed fine products that really altered my impression of "Made in China". Unitree's legged robot was among the world's best. Of course, this opinion may change once I get the opportunity to lay my hands on SpotMini. |
ROSCon at Macau -- 10.2019
|
ROSCon was very application-heavy with plethora of implementation-tips and know-hows of using ROS and ROS2.0.
My main motive for participating was scoping out ROS2.0. Humanoid platforms require precise timing in control due to legged-stabilization requirements. This means real-time performance and deterministic communication is critical -- something that was missing from previous version of ROS. I presented on developing a ROS API for humanoid platforms, and briefly discussed the challenges in motion generation & control problems in ROS. To recap my experience in ROSCon, ROS2.0 definitely offers an invaluable deterministic performance, which has led to many participation of industry also adopting ROS2.0. ROS2.0 even provides ROSbridge so that old-users can interface ROS2.0 as they did in previous versions. However, it is STILL in beta-stage with much custom development required. Firms reported almost 80% time was spend on developing ROS2.0 wrapper than actual related-work implementation. Unless I join a start-up and need this early-edge that other firms do not yet have, I don't think I'll be transitioning to ROS2.0 just yet. |
|
RoboWorld Exhibition @ S.Korea -- 10.2018
|
RoboWorld is the biggest Annual Robotic Exhibition in South Korea. KAIST Hubo Lab and Rainbow Robotics (c), both directed by Professor Oh, shared one of the center stage booths to demo our products. Our lab displayed total of 4 robots (humanoid, tele-operated robot, quadruped, and service robot) and Rainbow Robotics displayed their new RB5 collaborative robot arm.
This annual exhibition was a chance for our newly created service robot to shine. With new hardware system as well as software framework for robot intelligence, we successfully demonstrated autonomous drink delivery at relatively fast speeds to keep the audience entertained. |
|
Snowboard Trip using GoPro6 from Tooravel -- 01.2018
After completing my studies in Cornell (in New York), I have moved to work in Korea as a robotics researcher at KAIST.
After living in Korea for a year, I've experienced that Korea is a land that truly excels in providing service. More so than Samsung's phone technology or the famous K-Pop group, the variety of services that improve the quality of my day to day life here is incredible -- simply incomparable to the service I've become so used to in America.
After living in Korea for a year, I've experienced that Korea is a land that truly excels in providing service. More so than Samsung's phone technology or the famous K-Pop group, the variety of services that improve the quality of my day to day life here is incredible -- simply incomparable to the service I've become so used to in America.
|
One great example is Tooravel, a start-up that provides services to help prepare for a vacation, proving packages and even renting out a GoPro to capture special moments.
For less than $10/day, Tooravel mailed the GoPro6 to me with all the accessories I requested. For free, it included the 2 batteries, charger, 128GB SD card, basic mounts, and small fee for the additional head strap and 3way. It's surprising how much positive effect GoPro6 had on my snowboard trip! Because of the GoPro, my friends and I were incredibly active and stayed together as a group more. Instead of just enjoying the slope, we tried creative ways to capture the moment. We rode more, laughed more, and talked more. Renting from Tooravel was definitely worth the money, and I am excited to capture another trip next time! *To check-out Tooravel's site: http://www.tooravel.net/product/?product_id=1 |
|
Korea Robotics Society Annual Conference -- 01.2018
|
My very first conference experience was at KROS2018 in Korea. I presented on Vision-based Footstep Planner over Uneven Terrain of Humanoid Robot HUBO. To summarize, I created a 3D map for HUBO to detect uneven surfaces so it could walk over things.
KROS conference was an opportunity for me to share with others what I've been working on, and also hear about other researchers' solution to similar problems. To share knowledge but also to see that other people were also working in this field was very encouraging. 2018 goal is to attend an international conference! |
|
Glimpse of Robot History at the MIT Museum -- 06.22.2016
Me mimicking the robotic expression of Kismet
|
As my iRobot internship came to an end in Boston, I visited the MIT Museum to catch a glimpse of the development of robotics. I got to see surgical robots, various robotic leg dynamics, and socially intelligent robots. One thing that immediately stood out to me was the difficulty of prototyping without the aid of 3D printing. To know that everything, from the very early stage to the final design, had to be done with solid aluminum frames seemed a bit limiting to the development cycle. In addition to the outer frames, most of the electrical systems were done on perfboards, which would have added headache for placement, wire routing, and debugging. I felt almost privileged to have the aid of free and easy CAD software, access to manufacturing tools, and open-source resources.
|
Visiting the DARPA DRC Champion -- 07.14.2015
|
Only a couple weeks after the DARPA Robotics Challenge (DRC) 2015 finished, I got a chance to visit the winning team and robot in KAIST University, South Korea. 23 participating teams in DRC had to complete a timed obstacle course that comprised of 8 different tasks such as opening a door, climbing stairs and debris piles, operating a valve, and even driving a car. Seeing how couple other top teams were also able to complete all 8 tasks, HUBO's great advantage seems stem from its adaptable rolling-leg design that grants better stabilization and as a result, speed, in completing complex manipulation tasks. Seeing how recovering from a fall is a significantly complex task (only 1 robot from CMU recovered), the winning strategy seemed to be minimizing the probability of a fatal fall that could end the mission -- and the HUBO's rolling-leg achieved that.
To hear that HUBO was all custom-built in-house was astonishing. As an another Korea engineer interested in robotics, I was inspired to learn more and rethink the capabilities of humanoid robots in complex tasks. |
Tour of HUBO Lab with a researcher friend at KAIST
|
|
The video on the right is one of my favorite videos I saw from DRC 2015. While comical, it also helped me realize the insane difficulty to maintain stable body while performing manipulation tasks.
|
|
