Neural Field Dynamics Model for Granular Object Piles Manipulation. (arXiv:2311.00802v1 [cs.RO])

We present a learning-based dynamics model for granular material
manipulation. Inspired by the Eulerian approach commonly used in fluid
dynamics, our method adopts a fully convolutional neural network that operates
on a density field-based representation of object piles and pushers, allowing
it to exploit the spatial locality of inter-object interactions as well as the
translation equivariance through convolution operations. Furthermore, our
differentiable action rendering module makes the model fully differentiable and
can be directly integrated with a gradient-based trajectory optimization …

arxiv convolutional neural network dynamics exploit fluid dynamics manipulation material network neural network representation spatial