SPNets: Differentiable Fluid Dynamics for Deep Neural Networks
This work addresses the challenge of incorporating physical simulations into machine learning for robotics or graphics, though it appears incremental as it builds on existing neural network and fluid dynamics concepts.
The paper tackles the problem of integrating fluid dynamics with deep neural networks by introducing Smooth Particle Networks (SPNets), which enable differentiable fluid simulations and allow learning fluid parameters from data, achieving successful applications in liquid control tasks.
In this paper we introduce Smooth Particle Networks (SPNets), a framework for integrating fluid dynamics with deep networks. SPNets adds two new layers to the neural network toolbox: ConvSP and ConvSDF, which enable computing physical interactions with unordered particle sets. We use these lay- ers in combination with standard neural network layers to directly implement fluid dynamics inside a deep network, where the parameters of the network are the fluid parameters themselves (e.g., viscosity, cohesion, etc.). Because SPNets are imple- mented as a neural network, the resulting fluid dynamics are fully differentiable. We then show how this can be successfully used to learn fluid parameters from data, perform liquid control tasks, and learn policies to manipulate liquids.