Slippage Detection

Real-time grasp stability is based on successful slippage detection. In this work, we consider slippage detection as a binary problem (slip, stable) and we propose a novel set of temporal and frequential features, extracted from force norm profiles and collected during reliable ground truth labeling processes, finally employed within a classification framework.

More information can be found in:

I. Agriomallos, S. Doltsinis, I. Mitsioni and Z. Doulgeri, "Slippage Detection Generalizing to Grasping of Unknown Objects Using Machine Learning With Novel Features," in IEEE Robotics and Automation Letters, vol. 3, no. 2, pp. 942-948, April 2018. [link to the paper]