Above: Sid Misra (MBA 2013) cofounder, Perceptive Automata (Len Rubenstein; courtesy of Perceptive Automata)

Amid the hubbub of this year’s Consumer Electronics Show (CES), the world’s largest technology tradeshow, Sid Misra (MBA 2013) guided an onlooker to a leather driver’s seat positioned in front of a faux dashboard. A large monitor simulated the view through a windshield as a car drove on a quiet suburban street. On one side of the road, a woman walked casually down the bike lane. On the other, a young boy with a backpack ambled along the sidewalk—and then suddenly dashed in front of the car.

The human observer was not surprised by the child in the street. Subtle behavioral cues had told him the boy was planning to cross to his mother. What was surprising was that the self-driving car simulator recognized those cues, too. On the screen, the simulator tracked the motions of the pedestrians, correctly assigning an intention score (Do they want to cross the street?) and an awareness score (Do they know there is a car?) to each. Guided by technology developed by Perceptive Automata, the company cofounded by Misra, a self-driving car in this situation would hit the brakes.

“Self-driving cars today know what they are seeing and how fast the object is moving, but they aren’t good at predicting what a human will do next,” explains Misra. “We want to help machines understand humans.”

For Perceptive Automata, the road to CES (where the company exhibited in a prime spot with its partner, Honda) began in August 2014, when Misra started his year as a Blavatnik Fellow in Life Science Entrepreneurship, a program launched in 2013 at HBS. Among the numerous scientists Misra met with in his search for marketable life science ventures was an interdisciplinary group of computer vision, neuroscience, and behavioral science researchers. They had the tools to train computers to think more like humans, but did not yet have an application for the technology. Together the team—including David Cox, the John L. Loeb Associate Professor of the Natural Sciences and of Engineering and Applied Sciences at Harvard; Harvard fellow Walter Scheirer; and PhD candidate Sam Anthony, all cofounders—narrowed their focus to autonomous cars.

By 2016, the startup had a bank of desks in the Harvard i-lab. The cofounders were grateful for the access to the Harvard and HBS networks. It was through its affiliation with the i-lab that Perceptive Automata attracted its cofounder, Avery Faller, a machine learning engineer who was completing his master’s studies in computational science and engineering at Harvard. “He was the final piece of the puzzle,” says Misra. “And he found us through a spreadsheet of startups that the i-lab created.”

Today, Boston-based Perceptive Automata is working with suppliers and automakers in the United States, Europe, and Asia, and the technology is in testing in autonomous vehicles and highly automated human-driven vehicles.

The number-one question Misra gets—at CES and elsewhere—is “‘When will this be real?’ I tell them, ‘It’s real today.’” There are autonomous vehicles on the road, he notes. They aren’t driving through traffic in New York or Boston yet, but they will be. And when they do, Misra says, “We want them to be able to predict the actions of pedestrians, cyclists, and other motorists just like a human driver would.”

Post a Comment