Aaron T. Becker, assistant professor of electrical and computer engineering at the UH Cullen College of Engineering, earned a five-year, $550,000 CAREER award from the National Science Foundation (NSF) for his robotic research proposal titled, “Massive Uniform Manipulation: Algorithmic and Control Theoretic Foundations for Large Populations of Simple Robots Controlled by Uniform Inputs.” The NSF Faculty Early Career Development Program awards 600 grants each year to help promising young faculty members lay the foundations for successful academic careers.
In the 2014 Disney movie “Big Hero 6,” the protagonist, Hiro, offers a profound view into the future by manufacturing a swarm of 105 micro-robots. Hiro controls them to self-assemble, to build structures and to transport goods and materials. While the micro-robots of the film are fantasy, the ideas are rooted in reality. Producing large numbers of micro- and nano-robots is possible today. Micro-robots can be manufactured in large numbers by MEMS processes. Also, biological agents such as bacteria and paramecium can be grown to achieve large swarms.
Becker’s vision is for large swarms of robots remotely guided through the human body to cure disease, heal tissue and prevent infection, and ex vivo to assemble structures in parallel. The biggest barrier to Becker’s vision is a lack of control techniques that can reliably exploit large populations despite incredible under-actuation.
“Robotic manipulation at micro- and nano-scales can fundamentally transform how we treat diseases and assemble objects,” Becker said. “My goal is to precisely deliver materials and assemble structures from the bottom up.”
This precision manipulation must be coupled with a large population of manipulators to enable rapid progress. The potential impact is broad: large populations of micro-manipulators could provide targeted drug-delivery, perform minimally invasive surgery and engineer tissue.
Manipulation with these robots requires motion control. However, the small size of micro- and nano-robots severely limits computation, sensing and communication. Distributed control is infeasible – building autonomous robots is currently impractical at the micro-scale and seems impossible at the nano-scale. Instead, robots at this scale are currently powered by global force fields, such as a magnetic gradient or light broadcast at a specific frequency. Centralized approaches are feasible, but individually controlling a million robots requires an equally large amount of communication bandwidth, ultimately limiting the population size. Becker is designing new techniques for centralized control under the constraint that every robot receives exactly the same input commands. The unifying theme is using obstacles to efficiently control the shape, arrangement and position of the swarm.
In a drug-delivery application, blood vessels and other lumens serve as natural passageways to every part of the human body, so theoretically, external control algorithms could steer concentrations of drug particles to precise targets for more effective treatment of diseases.
Such a technique could revolutionize chemotherapy. Current treatment regimes flood cancer patients’ veins with toxins carefully calibrated to kill fast-growing cells. This targets tumors, but unfortunately also destroys cells that form hair and fingernails. With the controllers Becker is designing, physicians could use the body’s passageways to deliver concentrations of drugs with higher toxicity to specific areas with fewer patient side effects.
Furthermore, current robotic micro-assembly techniques use sophisticated micro-scale tweezers to individually place one component at a time. Becker’s lab is designing maze-like structures that, when actuated, simultaneously assemble multiple copies of a desired structure.
“In parallel, the process would look like a factory on a microchip,” Becker said.
Becker’s robotic research began using larger scale robots while he was a doctoral student at the University of Illinois at Urbana-Champaign. During his professional career, the scale of his work has gradually downsized with projects that involve 3-centimeters-tall kilobots programmed to move toward the brightest light source in the room and, more recently, magnetically-steered single-cell parameciums.
Becker’s initial algorithms could steer 100 robots at a time using the same global input – something like steering 100 remote-control cars with a single joystick. When Becker showed these algorithms to James M. Tour, a chemistry and nanoengineering professor at Rice University, Tour provided Becker with a bottle of 1014 (100,000,000,000,000) tiny synthetic molecules dubbed nanocars. These nanocars were invented in Tour’s lab and are arguably the smallest robots possible. When researchers shine light of a specific wavelength on the nanocars, the light activates a tiny molecular motor on each nanocar, propelling the cars forward. This light acts as the global control input, but the challenge is how to use this input to make the cars complete a desired task.
That challenge was the genesis of Becker’s work.
“I’ve been carrying around this bottle of nanocars in my backpack for the last two years, thinking about ways to push our algorithms to efficiently control incredibly large numbers of robots,” Becker said. “This has led to a growing list of algorithms and techniques that use the environment to manipulate the swarm.”
Many researchers are exploring ways to give intelligence to small swarms of robots to make them perform particular tasks in an environment. In contrast, Becker is looking for intelligent ways to exploit the environment to make large robot swarms perform particular tasks.
NSF CAREER awards require an emphasis on broader outreach that extends the research into society. Becker’s lab hosts externships with local HISD high school teachers, internships for a select group of high school students, and outreaches to local robotics clubs. He seeks to empower high school students and teachers, especially those from under-represented groups, to perform STEM outreach to younger students using rapid prototyping, intelligent online gamification and hands-on learning.
Becker developed a website, SwarmControl.net, that allows visitors to play games that compare and contrast several control theories for directing swarms of simulated robots. More than 10,000 people have participated in the two-year study thus far.
“SwarmControl.net gives us an ideal sandbox to test our theories. We have game challenges such as using a swarm of robots to push a ball through a maze,” Becker said. “For each game, we can test whether increasing the number of robots from 50 to 1,000 makes the challenge easier or harder, or test if it is better to have a swarm attracted or repelled by a user’s mouse click. Anyone interested in robots who has a computer can help us understand how best to control swarms.”
In his laboratory, Becker uses a swarm of 100 small robots to test control laws and algorithms. These robots are each about the size of a quarter. One experiment steers the robots to act as compliant manipulators that push around a slightly larger toy piano. These same control techniques will be implemented using global forces to make tiny particles perform useful tasks.
“The key insight is rather than steer individual particles, we treat the swarm as an entity,” Becker said. “We then push the swarm into walls to squeeze the swarm, collect the swarm into a dense mass or shape the swarm into a useful tool.”