l-dyno: framework to learn consistent visual features using robot's motion

Open in new window