Automated camera-exposure control for robust localization in varying illumination environments

The vision-based localization of robots operating in complex environments is challenging due to the varying dynamic illumination. This study aims to develop a novel automated camera-exposure control algorithm for illumination robust localization. First, a lightweight photometric calibration method is designed to model camera optical imaging. Based on the calibrated imaging, a photometric-sensitive image quality metric is developed for the robust estimation of image information. Then, a novel automated exposure control algorithm is designed to respond to environmental illumination variation rapidly and avoid visual degradation by utilizing a coarse-to-fine strategy to adjust the camera exposure time to optimum in real-time. Furthermore, a photometric compensation algorithm is introduced to fix the problem of the brightness inconsistency derived from the change of exposure time. Finally, several experiments are performed on the public datasets and our field robot. Results demonstrate that the proposed framework effectively improves the localization performance in varying illumination environments.