Real-Time Slip Detection using Tactile Information

Slip detection is of paramount importance for stabilized grasping of objects by a prosthetic hand. This paper presents a real-time slip detection framework using a data glove customized with force sensors. The data glove can acquire grasping force with a root mean square error (RMSE) of ±0.21 Newton. A finite state machine (FSA) algorithm was implemented for estimating the instances of slip occurrence as features. Support Vector Machine (SVM) with polynomial and radial basis function (RBF) kernel, k-nearest neighbor (k-NN), Naive Bayes (NB) and Random Forest algorithms were evaluated for detection of slip. An average accuracy of 94% and 98% was achieved using polynomial and RBF kernel SVM respectively. Further NB, k-NN and Random Forest algorithms resulted into an average accuracy of 96 %, 99 % and 100 % respectively. These experimental results show that the proposed framework is very useful for slip detection using tactile force information. It demonstrated robustness of FSA with machine learning algorithms for real-time slip detection and thereby holds promise for stabilized grasping by a prosthetic hand.