Slipping detection and avoidance based on Kalman filter

• A dynamic model is setup to test algorithms for friction estimation, slipping detection and avoidance.
• The slipping detection algorithms is based on a Kalman filter method.
• An haptic exploration is proposed to estimate unknown friction coefficient.
• The slipping avoidance control strategy works in both static and dynamic conditions.
• Experimental validation of presented methods is performed on a mechatronic system.

The purpose of this paper is to present the latest slipping detection and avoidance algorithms developed by the authors for application in robotic manipulation tasks. Slipping can happen not only in quasi-static conditions such as in grasping tasks but also during dynamic manipulation, therefore the availability of slip control techniques effective in both conditions, such as those proposed here, are essential in real robotic applications. A new algorithm is also proposed to estimate on-line the actual friction coefficient at the contact with the manipulated object by means of a preliminary exploration phase, thus enabling safe manipulation of objects with unknown surface properties. A detailed dynamic simulator is presented and experimentally validated on a mechatronic test bench used for proving the effectiveness of the proposed approach.