CMU prof says computers that can ‘see’ soon will permeate our lives
A pioneer in the science of making computers and robots see said self-driving cars are just the beginning.
Takeo Kanade, a professor at Carnegie Mellon University’s Robotics Institute and the recipient of the prestigious Kyoto Prize, said as cameras improve, as computer processors get faster and more mobile, and as connectivity increases, computer vision will begin to work its way into every part of life.
“We are at the verge of making this technology near ordinary life,” Kanade told the Tribune-Review via phone from San Diego, where he spoke as part of San Diego State University’s Kyoto Symposium.
But Kanade doesn’t envision a world in which robots take over all the work performed by humans. Instead, robots will help humans do what humans want to do but can’t. Kanade’s “magical equation for robots” is:
What robots should do equals What people want to do minus What people cannot do, plus or minus Delta.
Robots and computer vision will help people see, walk and communicate when they could not otherwise do so.
“In the past, robots have reduced people’s involvement,” Kanade said. “I think robots should help people be involved, because people want to do what they can do by themselves.”
Kanade, 71, spends much of his time in Japan but still conducts research for CMU. He was awarded the Kyoto Prize in Advanced Technology in November.
The Kyoto Prize, created in 1984 by Japan’s nonprofit Inamori Foundation, is the nation’s highest private honor for those who contribute to humankind’s scientific, cultural and spiritual development. Kyoto Prize winners receive a gold medal and 50 million yen, about $440,000. Several Kyoto Prize honorees also have won Nobel prizes.
Kanade’s work in computer vision started in the 1970s, when even basic digital images were rare. At that time, it was hard to anticipate where advances in cameras and computing power would go, but Kanade had a suspicion.
“I had a personal belief that computers should become as smart as humans, no questions about that,” Kanade said.
In many ways, Kanade’s research has made computers smarter than humans.
In the 1980s, when Uber’s Travis Kalanick and Tesla’s Elon Musk were children, Kanade was working on self-driving cars. He developed the first artificial intelligence capable of sensing freeway lanes, obstacles and other vehicles and changing lanes on its own. He and a team of researchers at CMU drove from Pittsburgh to San Diego in 1995 without touching the steering wheel in the “No Hands Across America” demonstration of autonomous vehicles.
Kanade’s technology went to the Super Bowl in 2001, when CBS debuted EyeVision. Viewers could see replays stitched together from 30 cameras. The image could rotate around the action.
Most recently, Kanade developed vehicle headlights that allow humans to see through rain more easily by reducing glare.
Kanade expects computer vision to spur advancements in medicine. Highly detailed computer readings of CT scans and MRIs can help diagnose a patient. Kanade has shifted his research to using computer vision to track how cells move, divide and die.
“And that is the basis for the advancement of biology,” he said.
He said computer vision can help people understand one another better. Computers can pick up on signals in the environment that humans miss. They can read changes in blood flow in parts of the human body.
“The cameras themselves are getting clever,” Kanade said. “The challenging area is to actually recognize and understand what you’re thinking and what you’re feeling, and that will help people live a better a life.”
Aaron Aupperlee is a Tribune-Review staff writer. Reach him at [email protected] or 412-336-8448.