DYNAMO: Assembly Motion Prediction

Dependency-Aware Deep Learning Framework for Articulated Assembly Motion Prediction. We introduce a model that predicts physically valid assembly motions by explicitly reasoning about part dependencies and kinematic constraints.

PowVRtool: Realistic Haptics

A handheld haptic device for realistic power tool feedback in VR-based manufacturing training. We model the physical torques and vibrations of real tools to provide immersive force feedback for training simulations.

Interacting Objects Dataset

A Dataset of Object-Object Interactions for Richer Dynamic Scene Representations. Capturing the complex physics of objects colliding, sliding, and stacking to benchmark dynamic scene understanding models.