Maurizio Monge

Papers

Total Citations

H-Index

About

Maurizio Monge is a leading researcher at the intersection of robotics, computer vision, and differentiable optimization. His primary research focuses on developing principled frameworks that bridge classical nonlinear optimization with modern deep learning, enabling end-to-end structured learning for perception and control systems. Monge’s most impactful contribution is the creation of **Theseus**, an open-source library for differentiable nonlinear least squares (DNLS) optimization built on PyTorch. This work, published in 2022 and accumulating 47 citations, provides a general-purpose, application-agnostic tool that allows robotics and vision systems to learn optimization-based models end-to-end, overcoming the limitations of prior application-specific implementations. By making DNLS accessible and efficient, Monge’s work has empowered researchers to integrate rich geometric and physical constraints directly into learning pipelines, advancing state-of-the-art in areas like SLAM, motion planning, and 3D reconstruction. His contributions are notable for their practical impact, providing a foundational infrastructure that accelerates research and development in structured learning.