Justin Carpentier

Inria, École normale supérieure, Paris, France.

Justin with Uarior, a puppy training to become a guide dog
With Uarior, a very brave puppy learning to become a guide dog.

I am a research director at Inria, where I lead the Willow research group. Willow is a joint team of Inria, École normale supérieure, and CNRS, based in the Department of Computer Science at ENS in Paris.

My research focuses on physical AI, with the aim of giving robots the ability to reason about and act in the physical world. My approach brings together perception, optimization, machine learning, simulation, and control to help robots understand their surroundings, anticipate the consequences of their actions, and interact with their environment. I develop the mathematical foundations and computational tools that support these capabilities, with an emphasis on translating theoretical advances into practical algorithms and open-source software.

In September 2024, I was awarded an ERC Starting Grant for ARTIFACT, The ARTIficial Motion FACTory, which began in September 2025. If you are interested in contributing to this research, please see our open positions or get in touch.

Biography

My path into robotics began with a background in computer science and applied mathematics. I graduated with highest honors from École normale supérieure Paris-Saclay in 2013. In 2014, I worked with Katja Mombaur as a visiting student in the Optimization in Robotics and Biomechanics group at the University of Heidelberg, Germany.

From 2014 to 2017, I pursued my PhD in the Gepetto research group at LAAS-CNRS in Toulouse. My doctoral research examined the computational foundations of anthropomorphic locomotion: how we can understand human walking and translate that understanding into mathematical formulations for humanoid robot motion. This work established a thread that continues through my research today—using mathematics to understand movement and turn that understanding into algorithms.

After a postdoctoral appointment in Gepetto, I joined Willow as a postdoctoral researcher in September 2018. My research has since broadened from locomotion to the connections between perception, learning, simulation, and control, with physical interaction as a common focus.

Open positions

At Willow, we welcome talented postdoctoral researchers and engineers interested in tackling challenging problems in robotics. To express your interest, please complete our application form, and I will get back to you to discuss opportunities.

Research overview

In February 2023, I gave a talk to the IEEE RAS TC on Model-based Optimization for Robotics entitled Progress and Prospects in Optimisation for Learning and Control in Robotics. The talk offers an overview of our contributions to the field at that time.

Collaborators and team

I collaborate with Francis Bach, Adrien Taylor, Jean Ponce, Cordelia Schmid, Josef Sivic, and Ivan Laptev, alongside a team of talented and passionate researchers and engineers:

PhD students

Postdoctoral researchers

Engineers

Alumni

I have also had the pleasure of working with many passionate and inspiring postdoctoral researchers, PhD students, engineers, and visitors:

My earlier robotics collaborations include work with Jean-Paul Laumond, Nicolas Mansard, Olivier Stasse, Mehdi Bennalegue, Andrea Del Prete and Steve Tonneau.

Software

I am the lead developer of several open-source robotics software projects. I believe open-source software helps advance both research and society by making cutting-edge methods broadly accessible.

stack-of-tasks/pinocchio stack-of-tasks/pinocchio
coal-library/coal/ coal-library/coal/
Simple-Robotics/proxsuite Simple-Robotics/proxsuite
Simple-Robotics/pycppad Simple-Robotics/pycppad
Simple-Robotics/aligator Simple-Robotics/aligator
Simple-Robotics/simple Simple-Robotics/simple
stack-of-tasks/eigenpy stack-of-tasks/eigenpy
loco-3d/crocoddyl loco-3d/crocoddyl
jrl-umi3218/jrl-cmakemodules jrl-umi3218/jrl-cmakemodules