Mobile Augmented Reality Based Annotation System: A Cyber-Physical Human System

One goal of the Industry 4.0 initiative is to improve knowledge sharing among and within production sites. A fast and easy knowledge exchange can help to reduce costly down-times in factory environments. In the domain of automotive manufacturing, production line down-times cost in average about $1.3 million per hour. Saving seconds or minutes have a real business impact and the reduction of such downtime costs is of major interest. In this paper we describe MARBAS, a Mobile Augmented Reality based Annotation System, which supports production line experts during their maintenance tasks. We developed MARBAS as Cyber-Physical Human System that enables experts to annotate a virtual representation of a real world scene. MARBAS uses a mobile depth sensor that can be attached to smart phones or tablets in combination with Instant Tracking. Experts can share information using our proposed system. We believe that such an annotation system can excel current maintenance processes by accelerating them. To identify applicable mesh registration algorithms we conducted a practical simulation. We used a 6 axis joint-arm robot to evaluate 7 different ICP algorithms concerning time and accuracy. Our results show that PCL non-linear ICP offers best performance for our scenario. Additionally, we developed a vertical prototype using a mobile depth sensor in combination with a tablet. We could show the feasibility of our approach augmenting real world scenes with virtual information.