VR Driven Unsupervised Classification for Context Aware Human Robot Collaboration

Human behavior, despite its complexity, follows structured principles that, if understood, will result in more reliable and effective collaborative automation environments. Characterizing human behavior in collaborative automation systems based on understanding the underlying context allows for novel advances in robotic human behavior sensing, processing, and predicting. Here, virtual reality, through integration of HTC Vive Arena Pro Eye Bundle, Leap Motion, and Unity 3D game engine, is used for safe and secure data collection on humans’ movements and body language in human robot collaborative environments. This paper proposes an unsupervised classification framework through integration of dynamic time warping and k-means clustering algorithm to enable robotics agents to understand humans’ intentions based on their body movements. Results display that the proposed framework is capable of identifying underlying intentions with an average accuracy, recall, and precision of 85%, 73%, and 75%, respectively.