User Perception Model for Wearable Supervision Systems

Wearable supervision systems ease the deployment of advanced mobile solutions for the control of industrial plants. These systems are used to provide operators with the adequate information to perform the required manual operations on industrial plants. This paper presents the adaptation strategy developed to ensure that the operator perceives accurately the plant state relayed by a distant server despite the varying network conditions. The information provided to the user is mainly in the form of Augmented Reality video rendered in a Head Mounted Display. Using experimental subjective testing, the user video Quality of Perception is modeled to determine continuously the best encoding parameters values resulting from the compromise between the fluidity and the level of detail for real-time interaction with the plant. This model is then used by an adaptation scheme to reject output bitrate disturbances due to the variations of the spatial and temporal video content. The proposed approach adapts in real-time the parameters values of the video encoder to track a given reference bitrate.