Nonlinear SVM based anomaly detection for manipulator assembly task

Abstract

There is much attraction of automation of difficult assembly by robotic manipulator. However, robots in factory should be overseen by human workers in order to check whether task condition is anomaly or not. In order to reduce human cost, anomaly detection for assembly task is important. A task to tighten a screw as one of assembly tasks is focused on. In this paper, we propose a method to generate high confidence area in the map of features based on nonlinear support vector machine with Gaussian kernel. By proposed method, a robot system can reduce occasions to make mistake in recognition of task conditions.