RedShow RMA

1. Built a reinforcement-learning workflow for a six-motor wheel-legged robot with position-controlled legs and velocity-controlled wheels.
2. Combined identified physical parameters with an adaptation module instead of relying only on broad domain randomization.
3. Integrated encoder and IMU feedback at 200 Hz with a 50 Hz policy and a 30 Hz adaptation module.
4. Deployed through OpenCR, RS-485 leg communication, and CAN wheel communication, and demonstrated physical obstacle traversal.
5. Presented the team prototype with Hyeokjo Kwon and Seongwoo Bae.

Reported results are based on the project presentation and physical prototype demonstration; a larger quantitative evaluation was not conducted.