Physics is a continuum.
A world driven by forces.
Traditional simulators approximate how objects behave by hand-tuned parameters. Video models capture rich motion without the forces behind it.
We learn the continuous physical dynamics both leave out.
The next evolution of simulation.
Our research combines video, internal structure, CAD, material and manufacturing data with real-world force, contact and behaviour measurements.
We serve industrial deformables first.
Our model research centers on complex deformation and contact forces—the behaviors that make physical prediction difficult and useful.
Research and pilot enquiries →
Robotics
Learned simulation for policy training and evaluation, with models developed to predict how objects respond to manipulation and applied forces.
World models
Physical conditioning for learned predictions, connecting visible motion with the forces, materials and structures that produce it.
A frontier lab grounded in industrial reality.
Mode Maison works across computer vision, graphics, simulation, applied mathematics, material measurement, robotics and manufacturing.
We build the instruments, datasets, representations and models the problem requires. We work with existing robotics stacks rather than asking partners to replace them. We test ambitious research against the physical systems that industry needs to automate.
Contact us →
Supported by
Mode Maison is a member of NVIDIA Inception.
© 2026 Mode Maison