Autonomous Visual Target Tracking

Association - Carnegie Mellon University

Centered Iframe

Motivation:

Demonstration of enhanced tracking capabilities of dexterous manipulators using visuo-motor control.

Application:

The main objective of this project was to develop a system that allows a robotic manipulator to interact with dynamically changing environments. The primary functionality involves detecting, tracking, and engaging with randomly moving objects displayed on a screen. The system efficiently follows each object and handle them in sequence, showcasing the manipulator's precision and adaptability.

Approach:

  • Use Aruco markers for targets on a large display
  • Track the markers by Visual Servoing using a Robot EF-fixed camera
  • Verify tracking performance/register hit using a spotter system
  • Spotter system: a independent vision-based hit validation system
overview figure
Figure 1:System Architecture.