The goal of our study is to create highly intelligent and efficient computational systems to solve challenges associated with our complex physical world -- from the production of visually captivating physics-based animations, to the design of novel drones and robots, to the discovery of new 3D printable materials, and eventually, to the investigation of the hidden mechanics, and the alluring beauty, underpinning our Nature.
The beauty and complexity of our physical world are alluring. We create simulation algorithms to capture this beauty in a virtual world, for generating movies and animations, and investigate this complexity in a computational context, for the benefit of scientific research. We develop novel data structures, meshing algorithms, and numerical PDE solvers to simulate and explore various types of complex physical systems, ranging from fluids (e.g., flame, bubble, paint, and toothpaste), to biomemetic structures (e.g., human heart and bird beak), to 3D printable soft bodies and drones.
We are building the next-generation computational infrastructures to automate the process of material discovery, and more broadly, to virtualize the process of scientific experiments, by coupling predictive simulation, data generation, and artificial intelligence. Our goal is to enable AI to uncover the hidden complex mechanics and further to create novel designs of complex physical systems with outstanding performance. For example, we have devised an intelligent system by coupling high-fidelity elastic simulaiton and unsupervised learning to uncover families of microstructures with extremal properties that have never been discovered by human scientists.
3D Printable Robots
The nascent technology of 3D printing is revolutionizing the process of fabricating complex objects. We invent simulation-enhanced user interaction systems and topology optimization frameworks to support the efficient design and fabrication of various complex physical systems, such as a pentacopter, a microstructural gripper, and a soft swimmer. These designs and their fabrication outcomes exhibit extreme complexity, specific functionality, and are highly customized. By developing these tools, we aim to bridge the gaps among the fabrication hardware, simulation software, and user interfaces.