Hyundai Invests in Human Prediction Software Company, Perceptive Automata

Software from startup Perceptive Automata doesn't read minds, but it is designed to do the next best thing – anticipate human actions. That makes it extremely useful for autonomous vehicle applications. So useful, in fact, that Hyundai's venture and innovation branch, Hyundai CRADLE, decided to invest in the Massachusetts-based company.

Right now, most autonomous vehicles are overly cautious, slamming on the brakes at the mere sight of a pedestrian (or at least, what is perceived as such. Sometimes the offending article is nothing more than an inanimate object, such as a runaway trash can, but that's another story).

Perceptive Automata is working on a solution for the human apprehension currently faced by self-driving cars. While people can predict the behavior of other erratic individuals, at the moment, artificial intelligence can not.

And that's what Perceptive Automata wants to change. The company advertises itself as a provider of "human intuition for machines". Its website reads, "We are solving what is often referred to as the hardest problem for robotic systems: understanding human behavior to enable the symbiotic large-scale deployment of automated systems in human-dominated environments."

Hyundai's Center for Robotic-augmented Design in Living Experiences (CRADLE) is interested in the startup's technology for use in its future autonomous vehicles.

"One of the biggest hurdles facing autonomous vehicles is the inability to interpret the critical visual cues about human behavior that human drivers can effortlessly process," said John Suh, vice president of Hyundai CRADLE. "Perceptive Automata is giving the AV industry the tools to deploy autonomous vehicles that understand more like humans, creating a safer and smoother driving experience."

Perceptive Automata's technology uses deep learning to anticipate the actions of pedestrians, cyclists and other drivers. To accomplish this, actual humans have been hired to look at clips depicting driving situations involving other humans. They then give a response as to what they would do in the scenario. Afterward, that information is fed into a neural network used to teach artificial intelligence modules.

The result is that, hopefully, self-driving cars will know what to expect from erratic humans. That way, they can provide a smooth, brisk ride instead of stopping for every person in sight.

"We are ecstatic to have an investor on board like Hyundai that understands the importance of the problem we are solving for self-driving cars and next-generation driver assistance systems," said Sid Misra, co-founder and CEO of Perceptive Automata. "Hyundai is one of the biggest automakers in the world and having them back our technology is incredibly validating."

Hyundai isn't the only company to take notice of Perceptive Automata's handiwork. Toyota is also backing the company, as is JAZZ Venture Partners and First Round Capital, Slow Ventures and others.

Earlier this month, the Perceptive Automata announced it had raised $16 million in Series A funding. Altogether, the investments bring the company's total capital to $20 million.

Sources: Bloomberg, Hyundai, Perceptive Automata