An accurate and efficient camera - based indoor positioning approach for intralogistic environments

Indoor positioning of forklift trucks is often employed to track and trace goods in a warehouse. In the industry camera-based systems, which track artificial reference points (markers) on the floor or at the ceiling, have gained increasing attention. These systems use algorithms inspired by augmented reality software to estimate the camera’s pose relative to a marker for 6 degrees of freedom. In this paper we show that in a simplified, yet realistic setup, where camera and marker plane are coplanar, 3 degrees of freedom are enough and devise an accurate, robust and highly computationally efficient algorithm to determine the relevant pose information. The advancements include position-based markertracking using a simple motion model, marker segmentation in the uncalibrated image and pose estimation using just the pinhole camera model. Conducted experiments show that both the algorithm’s performance and accuracy are superior to state-of-the-art pose estimation algorithms for this simplified setup.