Dongho Kang

Happy #NationalRoboticsWeek! Last year, we gave you a sneak peek at AthenaZero, our robotic manipulator built to tackle dynamic tasks like a human arm. Learn more about how this fast, precise robot can switch in an instant from a gentle touch to high force depending on what the task requires: rai-inst.com/resources/blog…

Can we teach legged robots to understand natural language and gestures? Check out @TaerimYoon_'s work to see how we are training robots to respond to verbal and physical commands, much like training a real dog. Preprint: arxiv.org/abs/2601.08422 Webpage: taerimyoon.me/Teaching-Robot…

It is remarkable to witness the strides we are making in humanoid robotics by leveraging human data and reinforcement learning. We are seeing significant breakthroughs in replicating behaviors that are not only natural and athletic but also fully steerable.

We’re excited to announce the product version of our Atlas® robot. This enterprise-grade humanoid robot offers impressive strength and range of motion, precise manipulation, and intelligent adaptability—designed to power the new industrial revolution. youtu.be/sd8ivhpjI6g

Visiting the Computer Science and Engineering department at my home uni, @SeoulNatlUni, for the first time in 10 years. I'll be on campus tomorrow (Tuesday) morning around 10:00 AM, thanks to a kind invitation from Prof. Jungdam Won. If you’re around campus, please ping me!

📄 Happy to share our recent RA-L paper "Whole-body Inverse Dynamics MPC for Legged Loco-Manipulation"! We introduces a whole-body MPC framework that unifies motion and force planning within a single control layer, which enables physically consistent, emergent behaviors that allow legged robots to tackle complex manipulation tasks. Congrats on Lukas Molnar on his amazing master thesis! Also thanks to all co-authors Gabriele Fadini, @eastskykang, Fatemeh Zargarbashi, @StelianCoros.

See Spot perform dynamic whole-body manipulation. Using a combination of reinforcement learning (RL) and sampling-based control, the robot is able to autonomously drag, roll, and stack tires weighing 15 kg (33 lb), well above its maximum arm lift capacity. Learn more about coordinating locomotion and manipulation processes: rai-inst.com/resources/blog…

Spot is playing Ping Pong! Spin is a crucial part of the game, but few robots can handle it. We show receiving and generating significant spin using MPC. Collaboration with David Nguyen and Zulfiqar Zaidi! Video: youtu.be/3GrnkxOeC14?si…. Paper: arxiv.org/pdf/2510.08754.

Again, a huge congratulations to @yarden_as for his excellent leadership on this project. I'm grateful for the opportunity to have contributed alongside such a brilliant team. Witnessing the impact of theoretical advances on practical robotics was a truly rewarding experience.

"SPiDR" offers an elegant solution for training safe policies for sim-to-real transfer without needing any real-world data, making it versatile and easy to integrate with existing pipelines. You can learn more at the project website: yardenas.github.io/spidr/

I'm thrilled to share "SPiDR: A Simple Approach for Zero-Shot Safety in Sim-to-Real Transfer" will be presented by Yarden As at #NeurIPS2025! I'm proud to have been part of the team behind this amazing project.

We also release some LaTeX sty and bib files used in the handbook. If you are writing an ICRA paper on SLAM, these should be useful. Visit our GitHub repo for details: github.com/SLAM-Handbook-…

We're organizing a workshop at humanoids about RL and model based control. Please submit an extended abstract for the poster session!!

@IntrovertInDis1 @ChongZitaZhang BD has been developing and refining this pipeline for many years, particularly since they began to commercialize their robots.

@IntrovertInDis1 @ChongZitaZhang For those choreographed movements with fine details and smoothness, pure reward shaping is never enough. You need a complete pipeline including motion animation synthesis tool, optimization procedure and motion tracking control.

BD is 2-3 years late to the party when others have almost lost interest in this topic.

@KyleMorgenstein @rai_inst Yes :)

@eastskykang @rai_inst Congrats! Which team, HALO?

After an incredible journey completing my PhD, I’m starting my next chapter as a research scientist at @rai_inst. I am excited to join the amazing team and contribute to the mission of creating the next generations of intelligent robots.

@ChongZitaZhang Well, we can make the same argument for humanoids. You could easily say that Boston Dynamics already did most of that stuff years ago. If being the first is the only thing that matters, there's really no point of doing that.

@ChongZitaZhang We're talking about a completely different set of goals and a different focus. Demos from Unitree are primarily about power, maneuverability and robustness. But creating the kind of elegant, symmetric motion you see in this demo... that's a whole different story.