Vision-Based Camera/Robot Pose Estimation Using Both Semantic and Geometric Features on LEGO Baseplates

We are exploring the possibility of using LEGO baseplate as artificial landmarks (ALs) for robots and cameras in calibration, navigation or Augment Reality (AR) applications. LEGO baseplates are rigid, widely-available, low-cost, and precisely-manufactured and appear to be great candidate for ALs. However, they are also monochromatic with low contrast and easily affected by lighting. To overcome those issues, we utilize geometric and semantic information in our algorithm design by leveraging grid pattern, circle stud shapes, and text patterns. Our algorithm has extensively utilized the information for cross validation in noise filtering and position refinement using robust estimation methods. We have implemented and successfully tested our algorithm. The results show that our algorithm can recover more than 95% stud centers as feature points which ensures pose estimation accuracy. Our experiments also show that LEGO baseplate produces significantly more accurate camera pose estimation results than that of existing state-of-the-art counterpart when both methods are deployed by users with no computer vision background.