Adrien Taylor

Institut national de recherche en sciences et technologies du numérique, Département d'Informatique, École Normale Supérieure

Papers

Total Citations

H-Index

About

Adrien Taylor is a leading researcher at the intersection of optimization, robotics, and machine learning, with a primary focus on developing fast, reliable numerical solvers for convex quadratic programming (QP). His major contributions center on bridging the gap between theoretical optimization and real-time robotic applications. Taylor is the architect behind the PROX-QP solver, a highly efficient and versatile QP solver designed to meet the stringent demands of robotics—from whole-body control and motion planning to state estimation. His foundational work, "PROX-QP: Yet Another Quadratic Programming Solver for Robotics and beyond" (2022, 54 citations), established a new standard for speed and robustness, while his subsequent 2025 paper (9 citations) further refined its real-time capabilities. Demonstrating his impact on machine learning, Taylor also introduced "QPLayer" (2023), a framework for the efficient differentiation of convex quadratic optimization layers within neural networks. This work enables the seamless integration of domain-specific optimization knowledge into deep learning architectures. With a growing citation footprint, Taylor is widely recognized for making advanced optimization practical for high-frequency robotic systems and differentiable programming.