In CRV 2018 [Demo Video]
This is a proof-of-concept project trying to show that the navigation can be done using only sparse semantic object features. With the help of recent advances in Deep Learning, state-of-the-art object detectors are light-, condition, viewpoint-invariant oppose to low-level traditional features such as SIFT, SURF, etc.
This is ROS-based implementation of our CRV 2018 paper on robust UAV visual teach and repeat using only sparse semantic object features. citation information:
@inproceedings{GhasemiShamshirdarVaughan_semantic_vtr,
Author = {Ghasemi, Amirmasoud and Shamshirdar, Faraz and Vaughan, Richard},
Title = {Robust UAV visual teach and repeat using only sparse semantic object features},
Booktitle = {Computer and Robot Vision ({CRV})},
Year = {2018}
}