Making “smart robotic matter” smarter

ECE Ph.D. student Danna Ma stands in front of a city skyline at night, holding her cat in her arms

What if you could create individual programmable robotic modules that could sense their environment and, depending on what they are sensing, engage in exploratory behavior, signal to each other, create structures with varying functions and abilities, and then self-disassemble and move on to further exploration and tasks?

The possibilities for such a system would be enormous.

One name given to this sort of collective of programmable robots is “smart robotic matter” and Danna Ma wants to make it even smarter by giving it the ability to engage in three-dimensional entanglement.

Ma is a doctoral student in Professor Kirstin Petersen’s Collective Embodied Intelligence Lab in Cornell’s School of Electrical and Computer Engineering (ECE). Because Ma’s undergraduate (National University of Singapore) and M.Eng. (Cornell ECE) degrees did not involve robotics, she spent her first two years in Petersen’s lab learning some of the foundational skills she would need to begin designing her own robotic systems.

“I told Kirstin to just throw me on any project where she felt some of my skills with nanofabrication and semiconductors would be most useful,” Ma said. “And then while doing that I learned how to fabricate soft sensors and how to use 3D printers and laser cutters.”

What Ma wanted was to immerse herself in the top to bottom design of every aspect of a modular robot system and it was two years before she had the skills and was ready to dive in.

Now in her fifth year, Ma has designed the Smarticle 2.0 from scratch, (based on a proof-of-concept designed in the lab of collaborator applied physics Professor Daniel I. Goldman at Georgia Tech.) Ma, Petersen, and their collaborators have put the robotic modules through their paces and Ma will present their results to the American Physical Society (APS) at their March 2023 meeting in Las Vegas.

Ma had many criteria to keep in mind at every step of the design process. For a system to be truly useful it should actuate, sense, and respond intelligently to the environment; it should have scalable fabrication and operation as well as coordination algorithms that ensure reliable autonomy; and ideally it should be low cost, low weight, low barrier-of-entry, and easy to use. Ma added one more criterion: the modules should be capable of 3D entanglement with each other to greatly increase the number and variety of shapes they could form and the functions they could fulfill as a collective.

Ma is excited to share her work and to take the next steps with this research, which includes stress tests to enable larger collectives and comparing observed behavior with behavior predicted in simulation. But before she can run the next experiments she needs to get busy building more robots. “My goal is to bring the scale up from ten to fifty or even one hundred robots.” Ma said.

If you are reading this in January or February of 2023, it is an excellent bet that you’ll be able to find Ma in the Collective Embodied Intelligence Lab, building what can only be called a swarm.

More Spotlights