Whether they’re inspired by a gold beetle to 3D print a multi-material electronic device, creating programmable soft 3D printing materials for robots and drones, or making it possible to watch 3D movies at home without having to wear 3D glasses, we’re always interested to hear what new innovation the researchers at MIT’s Computer Science and Artificial Intelligence Lab (CSAIL) are working on next. Currently, they’re focused on topology optimization for 3D printing, and developed a new design system that catalogs the physical properties of many tiny voxel clusters, which are the building blocks for larger 3D printable objects.
3D printers typically have a 600 dot per inch resolution, so designers have a lot of precise control over a 3D printed object’s microstructure and physical properties, such as strength and deformation. But it would take an insane amount of time to evaluate the physical effects of all possible combinations for an object made up of billions and billions of voxels, not to mention for a multi-material object. The new software system from CSAIL allows designers to exploit a 3D printer’s extremely high resolution, working with microscopic scale physical measurements while it evaluates macroscopic designs.
“Generative design is becoming more and more important for additive manufacturing,” CSAIL associated professor of electrical engineering and computer science Wojciech Matusik told 3DPrint.com. “For example, the Project Dreamcatcher by Autodesk provides very compelling demonstrations of generative design and its applications. In that category, topology optimization methods can automatically come up with designs that leverage the flexibility of 3D printers. Unfortunately, topology optimization approaches can only handle designs with a small number of elements. The approach we have develop can handle designs with extreme resolutions.”
Generative design such as that Autodesk is working on represents a major area of focus for the future of 3D printing and other advanced technologies; indeed, Autodesk has told us that generative design is key to the company’s future. Project Dreamcatcher has been in the works for some time now, and its applications are growing as it comes closer to release as Autodesk Generative Design. And MIT is among those taking notice of the potential and building on it.
Matusik and several MIT students – Mélina Skouras, a postdoc in Matusik’s group, CSAIL postdoc Bo Zhu, and Desai Chen, an electrical engineering and computer science graduate student – wrote a paper on their work, titled “Two-Scale Topology Optimization with Microstructures,” and presented at last week’s SIGGRAPH 2017 graphics conference. The SIMPLEX program of the US Defense Advanced Research Projects Agency (DARPA) supported their work.
“Conventionally, people design 3-D prints manually. But when you want to have some higher-level goal — for example, you want to design a chair with maximum stiffness or design some functional soft [robotic] gripper — then intuition or experience is maybe not enough. Topology optimization, which is the focus of our paper, incorporates the physics and simulation in the design loop,” explained Zhu, the first author on the paper. “The problem for current topology optimization is that there is a gap between the hardware capabilities and the software. Our algorithm fills that gap.”
The research team defined a space of physical properties, where a microstructure assumes a set location. Three standard measures make up a material’s stiffness and define its three-dimensional space:
- Describes a material’s deformation in the direction of an applied force; how far it can be stretched or compressed
- Describes a material’s deformation in directions which are perpendicular to an applied force; how much its sides bulge or contract when squeezed or stretched
- Measures a material’s response to shear, which causes different layers of the material to shift relative to each other
Any specific combination of these three measures defines a point in space, and the researchers used three differently-sized clusters for their experiments. Then, they randomly generated clusters that would combine a set of 3D printable materials in various ways, and used physics simulations to assess each cluster’s physical properties. In this way, the algorithm explores the whole space of properties and ends in a cloud of points, which “defines the space of printable clusters.”
Bernd Bickel, an assistant professor of computer science at the Institute of Science and Technology Austria, said, “The design and discovery of structures to produce materials and objects with exactly specified functional properties is central for a large number of applications where mechanical properties are important, such as in the automotive or aerospace industries. Due to the complexity of these structures, which, in the case of 3-D printing, can consist of more than a trillion material droplets, exploring them manually is absolutely intractable.”
The system was able to mathematically figure out if a cluster with a certain combination of properties was printable by calculating the level set function, which describes the shape the cloud of points takes. The last step requires the printable object to be optimized, using the team’s custom software, in order to get specifications of material properties up to hundreds of thousands of 3D printable clusters of voxels. While their database of the clusters that have been evaluated might not have exact matches for the projected specifications, it will have clusters with excellent approximations.
“The solution presented by Bo and colleagues addresses this problem in a very clever way, by reformulating it. Instead of working directly on the scale of individual droplets, they first precompute the behavior of small structures and put it in a database. Leveraging this knowledge, they can perform the actual optimization on a coarser level, allowing them to very efficiently generate high-resolution printable structures with more than a trillion elements, even with just a regular computer. This opens up exciting new avenues for designing and optimizing structures at a resolution that was out of reach so far,” said Bickel, who’s also head of the Institute of Science and Technology Austria’s Computer Graphics and Fabrication Group.
Tell us your thoughts in the CSAIL Topology Optimization forum thread at 3DPB.com.[Source: MIT CSAIL / Images: Computational Fabrication Group at MIT]