Deep Transfer Learning of Pick Points on Fabric for Robot Bed-Making

A fundamental challenge in manipulating fabric for clothes folding and textiles manufacturing is computing “pick points” to effectively modify the state of an uncertain manifold. We present a supervised deep transfer learning approach locate pick points using depth images invariance color texture. consider task bed-making, where robot sequentially grasps pulls at increase blanket coverage. perform physical experiments with two mobile manipulator robots, Toyota HSR Fetch, three blankets different colors textures. compare coverage results from (1) human supervision, (2) baseline picking uppermost point, (3) learned points. On quarter-scale twin bed, model trained combined data robots achieves 92% compared 83% 95% supervisors. The transfers novel 93% Average 193 beds suggest that transfer-invariant on can be learned.