Exhaustive Simulation Approach for a Virtual Camera Calibration Evaluation in Gazebo

Abstract

Camera calibration is an essential research field with a high potential for emerging algorithms and calibration targets. In this work, we present a virtual experiments approach in Gazebo simulator for exhaustive camera calibration methods and calibration targets evaluation. Key steps of the camera calibration workflow were adapted to the Gazebo simulation approach. We proposed a virtual camera calibration evaluation pipeline that includes camera modeling and calibration target’s pose generation in a viewing frustum. Experiments in the Gazebo demonstrated virtual environment feasibility for camera calibration evaluation while comparing checkerboard and circle grid targets, and allowed to achieve an acceleration of more than 30 times compared to the real-time. Experimental results exhibited a need for additional calibration steps incorporation, such as outliers rejection and optimal calibration target poses generation.