Visual Grounding of Learned Physical Models
This addresses the problem of physical reasoning and adaptation in AI for applications involving rigid objects, deformable materials, and fluids, representing a novel method for a known bottleneck.
The paper tackles the challenge of enabling computational models to perform physical reasoning and adapt to new environments by presenting a neural model that reasons about physics and makes future predictions based on visual and dynamics priors. The result shows that the model can infer physical properties within a few observations, allowing quick adaptation to unseen scenarios and accurate future predictions.
Humans intuitively recognize objects' physical properties and predict their motion, even when the objects are engaged in complicated interactions. The abilities to perform physical reasoning and to adapt to new environments, while intrinsic to humans, remain challenging to state-of-the-art computational models. In this work, we present a neural model that simultaneously reasons about physics and makes future predictions based on visual and dynamics priors. The visual prior predicts a particle-based representation of the system from visual observations. An inference module operates on those particles, predicting and refining estimates of particle locations, object states, and physical parameters, subject to the constraints imposed by the dynamics prior, which we refer to as visual grounding. We demonstrate the effectiveness of our method in environments involving rigid objects, deformable materials, and fluids. Experiments show that our model can infer the physical properties within a few observations, which allows the model to quickly adapt to unseen scenarios and make accurate predictions into the future.