Back Home
IROS 2016

Persistent Aerial Tracking System for UAVs

Matthias Müller, Gopal Sharma, Neil Smith, Bernard Ghanem

Persistent Aerial Tracking System for UAVs

Abstract

We present a persistent aerial tracking system for UAVs designed to maintain a continuous lock on a moving ground target over extended flight durations. Our approach combines a discriminative correlation filter for fast, robust per-frame tracking with a re-detection mechanism that recovers from target loss due to occlusion or field-of-view departure. The system runs in real-time onboard a UAV and is validated on diverse real-world sequences featuring partial occlusion, illumination change, scale variation, and fast target motion. Our system achieves reliable long-term tracking performance significantly exceeding that of existing aerial tracking methods.

Resources

Video

Citation

@inproceedings{mueller2016pat,
  title     = {Persistent Aerial Tracking System for {UAVs}},
  author    = {M{\"{u}}ller, Matthias and Sharma, Gopal and Smith, Neil and Ghanem, Bernard},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  year      = {2016}
}
Copied!