The task of navigating from one place to another is one that even simple life forms can easily accomplish, but computer scientists have long struggled to design robots that can navigate with anything close to the efficiency of a rat hunting for food, since it’s very hard to get a robot to find its way around in a cluttered and changing environment. For a rat foraging in an alleyway, however, the ability to remember its route and come home quickly is key to its survival.

A new collaborative project aims to use what we know about the street smarts of rats to help robots find their way. Funded by a grant from the U.S. Office of Naval Research, the project is being led by Michael Hasselmo, a professor in psychology and neuroscience, and associate director of the Center for Memory & Brain (CMB). He has spent years studying the navigational system in rats’ brains that allows them to remember their previous experiences in time and space.

The mammalian brain, it turns out, has developed a very efficient system for spatial memory, using specialized “grid cells” located in the brain’s entorhinal cortex. By monitoring their activity with electrodes, scientists have found that grid cells are constantly active as an animal navigates its environment, and the specific pattern of cell activation is repeated when a rat returns to a place it’s been before. Moreover, the cells activate in an organized way—some maintain their response as the animal moves several meters, others change their activity over just a few inches. Hasselmo explains that such a system makes sense because we remember spaces in different scales: sometimes we need to remember the layout of neighborhoods in a city, and other times we need to remember the layout of cubicles in an office building.

Recent work in his lab, led by graduate student Mark Brandon, suggests that these patterns arise from the interference of rhythmic oscillations in the electrical activity of brain cells. An ongoing collaboration with psychology, neuroscience, and CMB professors Howard Eichenbaum and Chantal Stern—who also directs the Cognitive Neuroimaging Laboratory—as well as researchers at MIT, the University of Texas at Austin, and University College London, will investigate whether a similar principle can be encoded in software to help robots remember their routes and positions as they move around. Hasselmo explains that robots usually represent space point by point, in a very detailed way. “It’s more efficient to talk about space on different scales,” he says, and breaking spaces up into a hierarchy of rooms, buildings, and neighborhoods “would make a robot better at communicating with humans about where it’s been.”