Sean Humbert and Shanelle Clarke.

Shanelle Clarke is co-leading research advancing robots and uncrewed aerial systems using the unlikeliest of models – insects.

Clarke, a postdoctoral scholar at the University of Colorado Boulder, is co-principal investigator of a $688,000 grant from the Air Force Office of Scientific Research into swarming behaviors of flying bugs.

The work has significant engineering applications.

“Nature is really efficient at coordination of flight in insects. Swarms can be on the order of thousands or millions, but they’re very good at not crashing into each other. Understanding the sensing and communications principles have commercial and defense uses,” Clarke said.

Clarke is working as a postdoctoral researcher in Sean Humbert’s lab. Humbert, a professor in the Paul M. Rady Department of Mechanical Engineering, is fellow co-principal investigator on the grant. 

His research focuses on unlocking bioinspired principles of insects and animals to advance engineering for society. Past studies have decoded how houseflies sense the environment around them and successfully navigate the world.

This new grant moves that research into the behavior of insect swarms.

“We’re taking optimization principles from the organism level to the swarm level. Insect swarms operate with more agility and robustness than synthetic/engineered counterparts, so any progress here will be a significant step forward,” Humbert said.

The grant will advance theoretical foundations and computational analysis of these behaviors.

“We want to be able to say which agents of the swarm are talking to each other, which should be leaders, which should be followers, and in moving through spaces, how they optimize directionality,” Clarke said.

A critical aspect of insect-based behavior at the individual or swarm level is how computationally streamlined it is. Flies are only 6-7 millimeters long and have brains the size of poppy seeds. In comparison, human built drones contain comparatively large computers and expensive sensors.

Current swarms of uncrewed aerial systems, like those used for light shows over stadiums, also use communication mechanisms that become increasingly problematic as the number of drones increases.

“How can we optimize those communications links? Right now, standard practice is that everybody is talking to everybody else, which creates communication bottlenecks. That’s not how insects do it. What can we learn from them?” Clarke said.

The swarming research is still early, but Humbert is excited at its potential.

“The mathematics underlying the optimization principles is quite elegant, and I’m excited to turn these tools loose on a new and extremely relevant set of problems,” he said.

The two-year grant covers development and tuning of computational algorithms for sensing and motion, and to scale up those tools to swarms with thousands of operators. The work will explore ideal swarm shapes, communications, how swarm leaders and subordinates should be selected, and optimization of movement.

“There’s some really cool math that’s going to come from this,” Clarke said. “To our knowledge, open-loop swarm co-design has never been done before. The nature inspired principles here are going to knock people’s socks off.”