ansh1113/terrain-aware-locomotion
Terrain-aware locomotion pipeline for quadruped robots - 95% navigation success, 50% fall reduction using depth camera perception, ML terrain classification, and adaptive footstep planning with MoveIt2
This pipeline helps quadruped robots like ANYmal C or Unitree A1 navigate challenging, real-world environments autonomously. It takes input from a robot's depth camera to understand its surroundings and outputs precise, adaptive footstep plans. Operations engineers, robotics technicians, or field deployment specialists for industrial inspection or logistics robots would use this.
Use this if you need quadruped robots to reliably move over diverse, unstructured terrains like stairs, slopes, or rubble, without frequently falling or getting stuck.
Not ideal if your robot operates exclusively on flat, predictable indoor surfaces or if you require extreme speed over precision and fall reduction.
Stars
8
Forks
1
Language
Python
License
MIT
Category
Last pushed
Mar 11, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/ansh1113/terrain-aware-locomotion"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
enactic/openarm
A fully open-source humanoid arm for physical AI research and deployment in contact-rich environments.
Shuijing725/awesome-robot-social-navigation
A curated list of robot social navigation.
thomashiemstra/fred
This my 3d printed robot arm project
jstmn/ikflow
Open source implementation to the paper "IKFlow: Generating Diverse Inverse Kinematics Solutions"
sizhe-li/neural-jacobian-field
Controlling diverse robots by inferring jacobian fields with deep networks! Let's make robots...