NeuralFluid: Neural Fluidic System Design and Control with Differentiable Simulation

TL;DR

A novel framework to explore neural control and design of complex fluidic systems with dynamic solid boundaries and a benchmark of design, control, and learning tasks on high-fidelity, high-resolution dynamic fluid environments that pose challenges for existing differentiable fluid simulators are presented.

Abstract

We present a novel framework to explore neural control and design of complex fluidic systems with dynamic solid boundaries. Our system features a fast differentiable Navier-Stokes solver with solid-fluid interface handling, a low-dimensional differentiable parametric geometry representation, a control-shape co-design algorithm, and gym-like simulation environments to facilitate various fluidic control design applications. Additionally, we present a benchmark of design, control, and learning tasks on high-fidelity, high-resolution dynamic fluid environments that pose challenges for existing differentiable fluid simulators. These tasks include designing the control of artificial hearts, identifying robotic end-effector shapes, and controlling a fluid gate. By seamlessly incorporating our differentiable fluid simulator into a learning framework, we demonstrate successful design, control, and learning results that surpass gradient-free solutions in these benchmark tasks.

Figures (9)

• Figure 1: Pipeline Overview. (1) Our pipeline starts with an initial parametric geometry and a neural network parameterized controller. (2) The fluid dynamics is then simulated using a dynamic Navier-Stokes solver. (3) The performance of the design and control is evaluated using a loss function, the gradients of which are then back-propagated through our end-to-end differentiable framework. (4) The gradient-based optimization iteratively improves the geometry and control to achieve the task goal. This pipeline allows for efficient geometry and control co-optimization.
• Figure 2: Tasks Overview. In each task, the blue dashed line represents the inlet, the red dashed line indicates the outlet, the white arrows show the flow direction, and the orange shapes and arrows denote the geometry and its motion direction.
• Figure 3: (a) Visualization of Amplifier. (b) Visualization of Flow Modulator.
• Figure 4: Artificial Heart.Left: visualization of the domain and the location of the muscles. Middle: Optimized control policy rollout visualization. Right: Optimization results visualization. The top and bottom diagrams visualize the cosine and the ECG target variants.
• Figure 5: Ablation Studies.Left: Optimization trajectories for Neural Heart with 7100 parameters under different initialization. Iterations are visualized on a log scale. Right: Log scaled loss-iteration curves of our gradient-based method and other gradient-free optimization methods.