Nathaniel Merrill

Papers

Total Citations

176

H-Index

About

Nathaniel Merrill is a leading researcher in robotics and autonomous systems, specializing in visual SLAM (Simultaneous Localization and Mapping), sensor fusion, and unsupervised deep learning for perception. His most impactful contribution is the development of a lightweight unsupervised deep neural network for visual loop closure detection (2018, 146 citations), which introduced a compact autoencoder-based architecture that dramatically improves the reliability and efficiency of large-scale real-time SLAM—a critical capability for long-duration autonomous navigation. Merrill also advanced the state of the art in dynamic environment perception through his work on Schmidt-EKF-based visual-inertial moving object tracking (2020, 19 citations), demonstrating that tightly-coupled estimation significantly outperforms decoupled approaches for joint localization and dynamic object pose tracking. Additionally, his versatile 3D multi-sensor fusion framework (2020, 11 citations) enables robust, lightweight 2D localization for low-cost autonomous ground vehicles, even under challenging conditions like poor calibration, low lighting, and dynamic obstacles. Collectively, his work bridges the gap between theoretical rigor and practical deployment, making autonomous systems more reliable, efficient, and accessible for educational and industrial applications.