Projects

Research Prototype · 2025

RedShow RMA

RedShow SYSID-A-RMA Sim-to-Real Project

An undergraduate research prototype on improving sim-to-real performance by combining system identification with an A-RMA-style reinforcement learning pipeline.

Reinforcement Learning Sim-to-Real System Identification IsaacLab
Highlight 1

Reduced sim-to-real gap with SysID + A-RMA

Highlight 2

Robust deployment via encoder / IMU feedback

Highlight 3

Undergraduate team prototype (3 members)

01

What I Did

  1. Focused on reducing domain gap while avoiding overly conservative domain randomization.
  2. Used encoder and IMU feedback, policy-rate control, and adaptation concepts for robust deployment.
  3. Presented as a team prototype with Hansol Park, Hyeokjo Kwon, and Seongwoo Bae.

Poster/PDF material exists locally; representative web media is pending.

02

Topics · Stack · Code

Focus

Reinforcement learning method for robust sim-to-real deployment

Topics

Reinforcement LearningSim-to-RealSystem IdentificationIsaacLab

Details

Track
Research Prototype
Stage
Undergraduate
Year
2025
Status
Past