Autonomous person-following system for Unitree Go1

Development of an autonomous person-following system for a quadruped robot, specifically the Unitree Go1.

Problem

The project addresses the growing need for advanced robotic assistance in various fields:

  • Guide dogs for the visually impaired
  • Security applications
  • Enhanced human-robot interaction

Approach

Integrated cutting-edge object detection, person re-identification, and navigation technologies to create a robust tracking and navigation system.

Hardware:

  • Unitree Go1 quadruped robot
  • Custom-designed backpack containing:
  • Jetson Orin Nano computer
  • ZED2i stereo camera
  • LiDAR sensor

This setup enables accurate perception and decision-making for autonomous navigation and person tracking.

Video

Team

  • Dario Dittli
  • Pablo Soler Garcia
  • Sam Sulaimanov

Course: Robot Learning, ETHZ, FS2024