Detecting Impulsive Movements to Increase Operators’ Safety in Manufacturing

Abstract

Interaction and collaboration of human and robot is becoming a quite common scenario, in manufacturing production. When an operator works with a robot, usually, controlled and predicable movements are used. However, the operator could execute unexpected impulsive movements, which may induce risks for his safety. The purpose of this methodological study is to identify such impulsive movements, by using an inertial sensor fixed on the wrist and by four features calculated on the gesture kinematics. The features were calculated on the measured linear acceleration and on the jerk using two epoch durations (100 ms and 250 ms). Then, features threshold that allow impulsive movements onset identification were evaluated. Results showed that the four features overall recognized impulsive movements during reaching task executed with different directions and cadences. Results showed also that calculating the features with shorter epoch duration gave better results compared to longer epoch duration. A combination of the four features calculated on short epoch duration would increase impulsive gesture detection, reducing occurrence of false positive. In conclusion, this methodology seems suitable to detect impulsive gestures and to intervene promptly with safety measures.