Stable Haptic Feedback Generation during Mid air Interactions using Hidden Markov Model based Motion Synthesis

Abstract

Generation of stable and realistic haptic feedback during mid-air gesture interactions has garnered significant research interests recently. But limitations of sensing technologies such as unstable tracking, non uniform sampling duration , occlusions during interactions etc distort motion based haptic feedback. In this paper, we propose an improved motion synthesis model which tracks human gestures during interaction and recreates smooth and synchronized motion data from detected Hidden Markov Model (HMM) states. The proposed model uses ideal motion data of humans and duration of HMM states recognised during gestures to modulate the realtime motion synthesis to synchronize with actual human motion speed. The synthesized and raw motion patterns were compared with ideal motion curve obtained from a benchmark motion capture system to verify the effectiveness of the proposed method. Subjective evaluation of virtual reality (VR) system based on synthesized motion data showed significant improvements in user satisfaction over raw motion based VR systems.