This repo contains the notes for the papers I have read & understood and the topics I have studied
- Convex_MPC: Dynamic Locomotion in the MIT Cheetah 3 Through Convex Model-Predictive Control [Paper]
- DribbleBot: Dynamic Legged Manipulation in the wild [Paper][Video][Code][Notes]
Read the pdf for hand written notes
-
Algorithm:Uses PPO to train policy
-
Hardware: Unitree Go1
-
Simulation: Issac GYM
-
Object Detection Module: YOLOv7 [Paper]
-
Limitations:
- Slow turning response
- Visual perception sensitivity to lighting
- Imprecision at high speeds
- Lack of geometry awareness (like steep slopes and staircases)