This video shows some highlights from a field test we conducted in June 2016 at an old sand and gravel pit in Sudbury, Ontario, Canada. We taught our robot a 5 km network of interconnected paths and then carried out 120 km of autonomous repeats on these paths using only stereo vision for feedback. The video shows some sections being repeated.
Our technique is called Visual Teach and Repeat 2, and is a significant advance over our earlier work in that (i) it uses a Multi-Experience Localization (MEL) technique to match live images to several previous experience of a path, and (ii) is able to do place-dependent terrain assessment to safeguard the robot and people around it, even in rough terrain with vegetation.
This video is associated with the following papers:
Paton M, MacTavish K A, Warren M, and Barfoot T D. “Bridging the Appearance Gap: Multi-Experience Localization for Long-Term Visual Teach and Repeat”. In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Daejeon, Korea, 9-14 October 2016
Berczi L P and Barfoot T D. “It’s Like Deja Vu All Over Again: Learning Place-Dependent Terrain Assessment for Visual Teach and Repeat”. In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Daejeon, Korea, 9-14 October 2016
