Human motion analysis using learning methods based on data collected from touch sensors and 3D cameras for Human – Robot collaborations
The aim of the study is to classify human palm movement using learning methods for human-robot interface applications. In the first step, information of human palm movements was collected using a 3D camera and EMG contact sensors. The information was expanded and enriched using augmentation techniques in order to have a sufficient size dataset for the training.
Different classification learning methods were examined.
The goal is to obtain an optimal classification by combining a classification made on information from the touch sensors and a classification made on depth images.
The various methods for classifying and integrating the information are examined and compared on both synthetic and realistic information.