BYU

Playing with Swarms

Hearing Dr. Michael Goodrich talk about his research in human-robot interaction, you might not realize that he works at all.

“[Human-robot interaction is] a great place to play,” Goodrich said. “It’s mostly play for me. . . . If it has a human and it has a robot, I am probably interested in it.”

Goodrich investigates the ways that humans and virtual or robotic swarms interact. Swarm structures, like those seen in bee or ant colonies, are incredibly powerful and dynamic.

“[In swarms], you’ve created this decentralized system where everybody [a robot, a bee, an ant, etc.] is making their own decisions and then they interact with each other,” Goodrich said. “Through this interaction, a collective behavior emerges which is really cool, even though all of the individual agents are kind of dumb.”

The emergent strengths of swarms also make them incredibly hardy.

“Swarms are really decentralized, and that means if you lose 10% of the robots, the system still works,” Goodrich said. “If you have a human controlling a system, or over-controlling a system, you introduce a single point of failure: you cut off the human, then nothing works. And the human doesn’t have perfect information about what is going on.”

But when recreated with robots or computers, swarms often have trouble playing nice with human operators.

“What we don’t want to do is have the human take control of all the agents­–bots, bees, whatever–in the collective, because then you lose the inherent resilience and robustness of this system,” Goodrich said. “We want to have a human to be able to nudge the colonies or the swarms to do things that are interesting.”

Goodrich’s current research focuses on creating effective interfaces between robots, digital swarms, and their human partners.

“We don’t know exactly how humans think, but we understand a lot about representation and intentionality,” Goodrich said. “We also understand a lot about how algorithms work, how bio-inspired algorithms work, how individual path planners work. They speak different languages, and the representations that they use are different. We try to create interfaces and interaction styles that bridge these two different representations and way of thinking.”

Goodrich and his students have already developed an interface which allows a police or military team composed of humans and robots to coordinate their actions. The three objectives of stealth, speed, and energy efficiency are each matched to a different color, like a painter’s palette, and the human team member can mix together “drops” of color in proportion to what is most important for the team. For example, if completing a task (like surrounding a suspect’s house) without being detected is the most important objective, the human team member can add three drops of the stealth color, two drops of the speed color, and one drop of the energy efficiency color. This allows the robot team member to understand the goals and means of a mission.

“We’re not trying to program the robots to think like humans, but we’re trying to create mechanisms which allow the human to understand what the robots are doing without the humans having to become experts at algorithms,” Goodrich said.

Goodrich is extremely optimistic about the future of robot and swarm technology, even if Hollywood often is not.

“I don’t anticipate Terminator or I, Robot kinds of things; the technology really isn’t that close,” Goodrich said. “But I can see robots doing household chores. I can see robots helping with elder care. I can see using robots in a war zone, finding where the cluster bombs that didn’t explode are or demining an area. I can see them used increasingly for search and rescue. I can see them in self-driving cars. I can see that these technologies are emerging. . . and I would like to have the interactions designed so that the robots and the humans can accomplish things together, where the robots are helping accomplish human-centered goals in fairly natural ways.”

While the robots of the future might be capable of incredible things, Goodrich is excited to be working in the field today.

“We’re kind of in a stage where chemistry was in the alchemy days,” Goodrich said. “We’re mixing chemicals, hoping that we don’t blow up stuff, hoping that gold emerges. A very exploratory phase.”

By Mitch Rogers Posted on