?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=RoboTAP%3A+Tracking+Arbitrary+Points+for+Few-Shot+Visual+Imitation&rft.creator=Vecerik%2C+Mel&rft.creator=Doersch%2C+Carl&rft.creator=Yang%2C+Yi&rft.creator=Davchev%2C+Todor&rft.creator=Aytar%2C+Yusuf&rft.creator=Zhou%2C+Guangyao&rft.creator=Hadsell%2C+Raia&rft.creator=Agapito%2C+Lourdes&rft.creator=Scholz%2C+Jon&rft.description=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%2C+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%2C+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%2C+stacking%2C+and+even+full+path-following+tasks+such+as+applying+glue+and+sticking+objects+together%2C+all+from+demonstrations+that+can+be+collected+in+minutes.&rft.subject=Training%2C+Visualization%2C+Tracking%2C+Stacking%2C+Production+facilities%2C+Planning%2C+Task+analysis&rft.publisher=IEEE&rft.date=2024-08-08&rft.type=Proceedings+paper&rft.language=eng&rft.source=+++++In%3A++Proceedings+-+IEEE+International+Conference+on+Robotics+and+Automation.++(pp.+pp.+5397-5403).++IEEE%3A+Yokohama%2C+Japan.+(2024)+++++&rft.format=application%2Fpdf&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10196975%2F1%2Frobotapicra24.pdf&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10196975%2F&rft.rights=open