TY  - GEN
A1  - Vecerik, Mel
A1  - Doersch, Carl
A1  - Yang, Yi
A1  - Davchev, Todor
A1  - Aytar, Yusuf
A1  - Zhou, Guangyao
A1  - Hadsell, Raia
A1  - Agapito, Lourdes
A1  - Scholz, Jon
PB  - IEEE
SP  - 5397
N2  - For robots to be useful outside labs and specialized factories we need a way to teach them new useful behaviors quickly. Current approaches lack either the generality to onboard new tasks without task-specific engineering, or else lack the data-efficiency to do so in an amount of time that enables practical use. In this work we explore dense tracking as a representational vehicle to allow faster and more general learning from demonstration. Our approach utilizes Track-Any-Point (TAP) models to isolate the relevant motion in a demonstration, and parameterize a low-level controller to reproduce this motion across changes in the scene configuration. We show this results in robust robot policies that can solve complex object-arrangement tasks such as shape-matching, stacking, and even full path-following tasks such as applying glue and sticking objects together, all from demonstrations that can be collected in minutes.
UR  - https://doi.org/10.1109/ICRA57147.2024.10611409
ID  - discovery10196975
T3  - IEEE International Conference on Robotics and Automation (ICRA)
Y1  - 2024/08/08/
KW  - Training
KW  -  Visualization
KW  -  Tracking
KW  -  Stacking
KW  -  Production facilities
KW  -  Planning
KW  -  Task analysis
N1  - This version is the author-accepted manuscript. For information on re-use, please refer to the publisher?s terms and conditions.
EP  - 5403
SN  - 1050-4729
TI  - RoboTAP: Tracking Arbitrary Points for Few-Shot Visual Imitation
CY  - Yokohama, Japan
AV  - public
ER  -