Optimizing Artificial Landmark Placement for Improving Indoor Localization of Mobile Robots

Abstract Precise localization in GPS-denied indoor environments can be achieved using artificial landmarks. However, the majority of existing artificial landmark techniques have a low detection range and have to be used in large numbers to allow for standalone pose estimation. This is often not possible in cluttered indoor spaces like factories. In this paper, we propose an analytical framework to determine the minimum number of artificial landmarks and their placement for a given environmental layout and robot path, for a maximum allowable uncertainty in robot pose, anywhere on the predefined path. The proposed approach is validated with an Extended Kalman Filter (EKF) based localization on a differential drive robot. The EKF implementation uses odometry and Inertial Measurement Unit (IMU) data for prediction and AprilTags as sparse artificial landmarks for correction. The proposed approach was validated by driving a mobile robot with EKF localization in different simulated and real-world scenarios. Based on the desired path to be followed by the robot, the minimum number and placement of artificial landmarks were determined for each scenario. The maximum uncertainty in the robot pose at any point on its path was less than the allowable maximum threshold in both simulated and real-world testing. While the validation is presented using AprilTags with an EKF localization framework, the proposed approach could be applied to other passive and active tags, beacons, or marker-based localization approaches.