Custom 18-DoF hexapod robot — Kuramoto-CPG locomotion controller, real-time gait visualiser, IMU heading-hold autonomous firmware, full mechanical CAD, and a 34-test verification suite.
-
Updated
Apr 29, 2026 - Python
Custom 18-DoF hexapod robot — Kuramoto-CPG locomotion controller, real-time gait visualiser, IMU heading-hold autonomous firmware, full mechanical CAD, and a 34-test verification suite.
Hybrid CPG + reinforcement learning gait control for a hexapod robot (Hiwonder SpiderPi) in NVIDIA Isaac Lab, with sim-to-real deployment. Published at ARIS 2026.
Unitree-Go1 bot trained on different Reinforcement Algorithms prominently on PPO, TD3 and SAC (all with CPG). The repo provides a baseline to compare different DL algorithms based on their convergence and performance per 10M steps. The postures used for these are Jumping and walking.
To associate your repository with the central-pattern-generator topic, visit your repo's landing page and select "manage topics."