Plant maintenance support system by augmented reality

We propose a color-video-based augmented reality (AR) system designed for maintenance crews of plant equipment. The system superimposes visual guidance and instruction (e.g., the procedures, real time values of the system parameters, and the plans) upon real video images of the devices. It is a completely stand-alone system. The user downloads the necessary data of the target device from the databases, and uses this system in front of the device. Instead of image recognition, the system calculates accurate alignment of annotation and the device in real time by calculating the camera pose based on the captured relative positions of three square green markings on the target device. We evaluate a prototype system.