ARAS-IREF: An Open-Source Low-Cost Framework for Pose Estimation

Abstract

Despite the amount of research reported on state estimation and sensor fusion in the field of robotics, there are no well known low-cost solutions for a referencing system to determine the accuracy of developed methods by providing a suitable ground truth for them. In this paper an efficient and accurate 6-DoF pose measurement system is proposed and implemented on a spherical parallel robot using IR LEDs. This approach uses the perspective-n-point algorithm to derive the transformation matrix representing the accurate relative pose of the end-effector with respect to an inertial frame. Exploiting a visible light filter in front of the camera has rendered this approach robust against illumination changes. Furthermore, it allows for mitigating the rolling shutter effects by reducing the exposure time. Finally, a custom made testing module is proposed to verify the accuracy of the proposed device, and the calibration process proves the accuracy and efficiency of the system.