Improving the perception of visual fiducial markers in the field using Adaptive Active Exposure Control

TL;DR

Underwater localization using visual fiducial markers is challenged by harsh lighting; this paper proposes Adaptive Active Exposure Control (AAEC) that optimizes exposure time around the fiducial by a gradient-based quality metric and updates exposure frame-by-frame with momentum-enhanced gradient ascent. A dynamic region-of-interest around the marker and a gradient-aware weighting scheme improve robustness to lighting changes and fast convergence. Extensive field and motion-tracking experiments show AAEC substantially improves localization precision (covariance determinant) and detection rates, especially under adversarial lighting, compared with baseline and global exposure methods. Integrating AAEC with visual fiducial sensing can elevate real-time state estimation and control for AUVs, and future work targets multi-target RoIs, down-sampled processing, and integration with visual odometry.

Abstract

Accurate localization is fundamental for autonomous underwater vehicles (AUVs) to carry out precise tasks, such as manipulation and construction. Vision-based solutions using fiducial marker are promising, but extremely challenging underwater because of harsh lighting condition underwater. This paper introduces a gradient-based active camera exposure control method to tackle sharp lighting variations during image acquisition, which can establish better foundation for subsequent image enhancement procedures. Considering a typical scenario for underwater operations where visual tags are used, we proposed several experiments comparing our method with other state-of-the-art exposure control method including Active Exposure Control (AEC) and Gradient-based Exposure Control (GEC). Results show a significant improvement in the accuracy of robot localization. This method is an important component that can be used in visual-based state estimation pipeline to improve the overall localization accuracy.

Figures (8)

• Figure 1: Default exposure control over exposes visual fiducial.
• Figure 2: Visual fiducial markers captured in the Connecticut River with global (b) and our proposed local (a) exposure optimization.
• Figure 3: The workflow of AAEC.
• Figure 4: (a) Testing rig used for field testing our approach in the Connecticut river. (b) Comparison of algorithms (1) default, (2) AEC, (3) GEC, (4) AAEC in river water.
• Figure 5: Measured positions for different exposure control methods for river field test.