Norman Marlier
Papers
2
Total Citations
4
H-Index
2
About
Norman Marlier is pioneering the frontier of dexterous robotic manipulation through the lens of Bayesian inference. His research focuses on solving the fundamental challenge of robotic grasping under uncertainty, particularly for multi-fingered hands in cluttered environments. Marlier’s major contribution lies in introducing simulation-based inference to robotic grasping, a paradigm shift that treats hand configuration planning as a probabilistic inference problem rather than a deterministic search. His 2021 work on simulation-based Bayesian inference for multi-fingered grasping directly addresses the rich, nonsmooth contact dynamics and sensor noise that have long hindered dexterous manipulation. Building on this foundation, his 2024 paper advances the field with sequential neural ratio estimation combined with neural implicit representations, enabling 6-DoF grasping that reasons about uncertainty in real-time. Though early in his career, Marlier’s work represents a conceptually elegant synthesis of modern probabilistic machine learning and classical robotics, offering a principled path toward universal picking. His approach promises to unlock the full potential of dexterous hands in manufacturing, logistics, and service robotics by replacing brittle heuristics with robust, uncertainty-aware decision-making.
Research Focus
Key Achievements
Top Papers
- 1
- 2Simulation-based Bayesian inference for multi-fingered robotic grasping2 citations · 2021