Three 91探花 faculty members have been awarded early-career fellowships from the Alfred P. Sloan Foundation. The new Sloan Fellows, announced Feb. 20, are and , both assistant professors in the Paul G. Allen School of Computer Science & Engineering, and , an assistant professor in the Department of Chemistry in the College of Arts & Sciences.

Sloan Fellowships are open to scholars in eight scientific and technical fields 鈥� chemistry, computer science, Earth system science, economics, mathematics, neuroscience and physics 鈥� and honor early-career researchers whose achievements mark them among the next generation of scientific leaders.

The 126 were selected by researchers and faculty in the scientific community. Candidates are nominated by their peers, and fellows are selected by independent panels of senior scholars based on each candidate鈥檚 research accomplishments, creativity and potential to become a leader in their field. Each fellow will receive $75,000 to apply toward research endeavors.

This year鈥檚 fellows come from 57 institutions across the United States and Canada, spanning fields from evolutionary biology to data science.

Du鈥檚 research interests are in the theoretical foundations of machine learning, such as deep learning, representation learning and reinforcement learning.

“Recent breakthroughs in machine learning have relied on large neural network models trained on big data. These powerful models have become the predominant method in many data-driven domains. Another direction of machine learning that is experiencing a paradigm shift is data-driven decision-making, witnessed by increasingly capable self-driving cars, and applications aiming to align with human values, such as in ChatGPT,鈥� Du said. “However, we still don’t have a good understanding of why these paradigms are so powerful. My research aims to open the black box by building the theoretical foundations of modern machine learning paradigms that involve large models and decision-making.”

厂肠丑耻濒锄鈥檚 research creates design tools and systems that aim to revolutionize how physical聽artifacts are built. A central challenge for design tools used in manufacturing is the need to聽simultaneously nurture the creative ability to conceive novel designs and the analytical capacity聽to critically evaluate and optimize functionality and production. In addition to increasing productivity and product quality, her work empowers people of diverse backgrounds to design and create. For example, she worked with 91探花Medicine to craft custom personal protective equipment at the onset of the COVID-19 pandemic.

鈥淢y research tackles the fundamental challenges in manufacturing-oriented design through innovative solutions that are grounded in the fundamentals of geometry processing and combine insights from machine learning and programming languages,鈥� Schulz said. 鈥淢oving forward, I plan to expand my efforts on sustainable design, exploring innovative design solutions that prioritize reusability and recyclability to foster circular ecosystems.鈥�

Velian鈥檚 research program targets the design of new materials that contribute to decarbonization, clean energy and quantum information technologies. A special focus of her program is to bring molecular precision into the synthesis of single-site catalysts that transform abundant molecules into compounds poised to play central roles in a green economy.

“The traditional trial-and-error approach, which has been effective in developing conventional industrially relevant catalysts such as the Haber-Bosch process for converting nitrogen聽to ammonia, falls short in addressing current urgent chemical challenges, such as transforming other small molecules like carbon dioxide into environmentally friendly, valuable compounds,” Velian said. “Synthetic strategies to precisely control the composition and surface chemistry of inorganic materials are necessary to design the next generation of catalytic materials.”

For more information, contact Du at ssdu@cs.washington.edu, Schulz at adriana@cs.washington.edu, and Velian at avelian@uw.edu.

At the beginning of the COVID-19 pandemic, car manufacturing companies such as Ford quickly .

To make this switch possible, these companies relied on people working on an assembly line. It would have been too challenging for a robot to make this transition because robots are tied to their usual tasks.

Theoretically, a robot could pick up almost anything if its grippers could be swapped out for each task. To keep costs down, these grippers could be passive, meaning grippers pick up objects without changing shape, similar to how the tongs on a forklift work.

A 91探花 team created a new tool that can design a 3D-printable passive gripper and calculate the best path to pick up an object. The team tested this system on a suite of 22 objects 鈥� including a 3D-printed bunny, a doorstop-shaped wedge, a tennis ball and a drill. The designed grippers and paths were successful for 20 of the objects. Two of these were the wedge and a pyramid shape with a curved keyhole. Both shapes are challenging for multiple types of grippers to pick up.

The team will Aug. 11 at SIGGRAPH 2022.

“We still produce most of our items with assembly lines, which are really great but also very rigid. The pandemic showed us that we need to have a way to easily repurpose these production lines,” said senior author , a 91探花assistant professor in the Paul G. Allen School of Computer Science & Engineering. “Our idea is to create custom tooling for these manufacturing lines. That gives us a very simple robot that can do one task with a specific gripper. And then when I change the task, I just replace the gripper.”

Passive grippers can’t adjust to fit the object they’re picking up, so traditionally, objects have been designed to match a specific gripper.

“The most successful passive gripper in the world is the tongs on a forklift. But the trade-off is that forklift tongs only work well with specific shapes, such as pallets, which means anything you want to grip needs to be on a pallet,” said co-author , 91探花assistant professor of mechanical engineering. “Here we’re saying ‘OK, we don’t want to predefine the geometry of the passive gripper.’ Instead, we want to take the geometry of any object and design a gripper.”

For any given object, there are many possibilities for what its gripper could look like. In addition, the gripper’s shape is linked to the path the robot arm takes to pick up the object. If designed incorrectly, a gripper could crash into the object en route to picking it up. To address this challenge, the researchers had a few key insights.

“The points where the gripper makes contact with the object are essential for maintaining the object’s stability in the grasp. We call this set of points the ‘grasp configuration,'” said lead author , who completed this research as a 91探花undergraduate student in the Allen School.聽 “Also, the gripper must contact the object at those given points, and the gripper must be a single solid object connecting the contact points to the robot arm. We can search for an insert trajectory that satisfies these requirements.”

When designing a new gripper and trajectory, the team starts by providing the computer with a 3D model of the object and its orientation in space 鈥� how it would be presented on a conveyor belt, for example.

“First our algorithm generates possible grasp configurations and ranks them based on stability and some other metrics,” Kodnongbua said. “Then it takes the best option and co-optimizes to find if an insert trajectory is possible. If it cannot find one, then it goes to the next grasp configuration on the list and tries to do the co-optimization again.”

Once the computer has found a good match, it outputs two sets of instructions: one for a 3D printer to create the gripper and one with the trajectory for the robot arm once the gripper is printed and attached.

The team chose a variety of objects to test the power of the method, including some from a that are the standard for testing a robot’s ability to do manipulation tasks.

“We also designed objects that would be challenging for traditional grasping robots, such as objects with very shallow angles or objects with internal grasping 鈥� where you have to pick them up with the insertion of a key,” said co-author , a 91探花doctoral student in the mechanical engineering department.

Play this video to see how a gripper can pick up one of the challenging shapes: a 3D-printed wedge. Credit: 91探花

The researchers performed 10 test pickups with 22 shapes. For 16 shapes, all 10 pickups were successful. While most shapes had at least one successful pickup, two did not. These failures resulted from issues with the 3D models of the objects that were given to the computer. For one 鈥� a bowl 鈥� the model described the sides of the bowl as thinner than they were. For the other 鈥� an object that looks like a cup with an egg-shaped handle 鈥� the model did not have its correct orientation.

The algorithm developed the same gripping strategies for similarly shaped objects, even without any human intervention. The researchers hope that this means they will be able to create passive grippers that could pick up a class of objects, instead of having to have a unique gripper for each object.

One limitation of this method is that passive grippers can’t be designed to pick up all objects. While it’s easier to pick up objects that vary in width or have protruding edges, objects with uniformly smooth surfaces, such as a water bottle or a box, are tough to grasp without any moving parts.

Still, the researchers were encouraged to see the algorithm do so well, especially with some of the more difficult shapes, such as a column with a keyhole at the top.

Play this video to see how a gripper can pick up the column with a keyhole at the top. Credit: 91探花

“The path that our algorithm came up with for that one is a rapid acceleration down to where it gets really close to the object. It looked like it was going to smash into the object, and I thought, ‘Oh no. What if we didn’t calibrate it right?'” said Good. “And then of course it gets incredibly close and then picks it up perfectly. It was this awe-inspiring moment, an extreme roller coaster of emotion.”

, who completed this research as a master’s student in the Allen School, is also a co-author on this paper. This research was funded by the National Science Foundation and a grant from the Murdock Charitable Trust. The team has also submitted a patent application: 63/339,284.

For more information, contact Kodnongbua at milink@cs.washington.edu, Lipton at jilipton@uw.edu, Schulz at adriana@cs.washington.edu, Good at iangood@uw.edu and Lou at louyu27@cs.washington.edu.

Grant number: EEC 2035717

As the holidays approach, people might be thinking of neat do-it-yourself woodworking projects to give as gifts. But there’s often a disconnect between designing an object and coming up with the best way to make it.

Now researchers at the 91探花 have created , a digital tool that allows users to design woodworking projects. Once a project is designed, the tool creates optimized fabrication instructions based on the materials and equipment a user has available. The team Nov. 19 at SIGGRAPH Asia in Brisbane, Australia.

“To make a good design, you need to think about how it will be made,” said senior author , an assistant professor in the Paul G. Allen School of Computer Science & Engineering. “Then we have this very difficult problem of optimizing the fabrication instructions while we are also optimizing the design. But if you think of both design and fabrication as programs, you can use methods from programming languages to solve problems in carpentry, which is really cool.”

For Carpentry Compiler, the researchers created a system called Hardware Extensible Languages for Manufacturing, or HELM. HELM is composed of two different programming languages: a high-level language for designing an object, and then a low-level language for the fabrication instructions.

“Say I want to make a piece of wood that’s cut at a 45-degree angle,” Schulz said. “In the design user interface, I create a box and then I draw a line where I want the cut to be and tell the computer ‘Remove this part.’ That’s the high-level language. Then the low-level language says ‘Take a two-by-four, take your chop saw, set up your chop saw for a 45-degree angle, align the lumber to your chop saw and chop.'”

As the user designs an object using the high-level language, which looks similar to standard CAD software, a compiler verifies that the design is possible based on what tools and materials the user has specified they have. Once the user is finished designing, the compiler comes up with a set of optimal fabrication instructions based on different costs.

“If you want to make a bookcase, it will give you multiple plans to make it,” Schulz said. “One might use less material. Another one might be more precise because it uses a more precise tool. And a third one is faster, but it uses more material. All these plans make the same bookcase, but they are not identical in terms of cost. These are examples of tradeoffs that a designer could explore.”

The compiler has to sift through a huge space of possible combinations of instructions to find the best ones. But if it treats fabrication instructions like a program, then it can use programming tricks to simplify its search and select promising candidates.

“One program might have a good way to make the edge of the table; another one finds a good way to make the legs,” said co-author , an associate professor in the Allen School. “And we can find those and recombine them to make the best overall plan.”

Currently Carpentry Compiler is optimizing fabrication plans based on fabrication time and precision. In the future, the team would like it to take into account grain orientation and uncertainty in using specific types of tools. From there, the team hopes to expand this idea to more complex projects 鈥� such as 聽a project that requires woodworking and 3D printing.

“The future of manufacturing is about being able to create diverse, customizable high-performing parts,” Schulz said. “Previous revolutions have been about productivity mostly. But now it’s about what we can make. And who can make it.”

Additional co-authors are , a doctoral student at Tsinghua University who completed this research as a visiting student at the UW; , a postdoctoral research associate in the Allen School; , a doctoral student in the Allen School; and , an assistant professor in the UW’s mechanical engineering department. This research was funded by the National Science Foundation, an Adobe Research fellowship and a Tsinghua scholarship for overseas graduate students.

For more information, contact Schulz at adriana@cs.washington.edu.

Grant numbers: 1813166, 1644558