Korean high school students train humanoid and quadruped robots in Isaac Sim for cube collection

Korean high school students' Isaac Sim robot competition — just wanted to share

Summary

A non-profit competition in Korea enabled high school students to train walking policies for Unitree H1 humanoids and Go2 quadrupeds using Isaac Sim. Participants tuned reward functions to navigate arenas with stairs, ramps, and moving cars despite having little prior reinforcement learning experience.

Source: ROS Discourse — Community updates · Read original article ↗

Article text · Original source · English

Hi everyone,

This isn’t a ROS project, it’s all Isaac Sim / Isaac Lab, but I wanted to share it anyway because the students worked really hard.

We run a non-profit robotics competition for high school students in Korea.

Each team trained walking policies for a humanoid (Unitree H1) or a quadruped (Go2) by tuning reward functions. In the finals, the robots collect cubes in an arena with stairs, ramps and a moving car. Most of the students had never done reinforcement learning before this year.

Training records (successes and failures): [robotics.connect.or.kr/researchlog](ì°êµ¬ ê¸°ë¡ | NCRC 2026)
Match videos: https://www.youtube.com/@ncrc_official

Please take a light look. Thanks!

What the source reports

Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.

paper: not_reported

Source excerpts and review record

Automatically extracted; no manual editorial approval recorded.

Most of the students had never done reinforcement learning before this year. Training records (successes and failures): [robotics.connect.or.kr/researchlog]( ì°êµ¬ ê¸°ë¡ | NCRC 2026 ) Match videos: https://www.youtube.com/@ncrc_official Please take a light look. Thanks!
Open source S3