MIT Researchers Make Info-Gathering Robots Work More Collaboratively for Major Efficiency Gains
By having a team of 10 robots work together in planning a search, up to 92 percent of the computational effort can be saved.
A team of MIT researchers have published a paper detailing an algorithm which helps information-gathering robots work cooperatively, instead of competitively — eliminating wasteful maneuvers.
"The assumption has been that it never hurts to collect more information," lead author Xiaoyi Cai explains of the current state-of-the-art in robotic information gathering. "If there’s a certain battery life, let’s just use it all to gain as much as possible. We want to be cognizant of the trade-off between information and energy. It’s not always good to have more robots moving around. It can actually be worse when you factor in the energy cost."
The solution is an algorithm which no longer assumes energy is there to waste, and instead aims to optimise the balance between energy expended and information gathered — and a key part of that is ensuring the robots are treated as a team, rather than a collection of individuals. "It’s more of a collaborative effort," Cai notes. "The robots come up with the team plan themselves."
"We allow the robots to plan their trajectories on their own. Only when they need to come up with the team plan, we let them negotiate. So, it’s a rather distributed computation. Our method provides comfort, because we know it will not fail, thanks to the algorithm’s worst-case performance."
According to experimentation, the new approach offers a 60 percent saving in communication and between 80 and 92 percent in computation based on a comparison to a naive distributed execution approach based on a team of ten robots.
The team's work is to appear in ICRA 2021, with a pre-print available under open access terms on arXiv.org.