LAS VEGAS — Nearly everywhere you turn at this year’s Consumer Electronics Show people are talking about self-driving cars.
But don’t believe the hype.
Much of the science to get autonomous cars safely on the road still needs to be accomplished. In other words, there’s a long way to go on this road trip.
That’s the perspective of Dr. Gill Pratt, the CEO of Toyota Research Institute, the car company’s $1 billion effort to transform the driving experience through artificial intelligence and machine learning.
“In truth, we are a long way from the finish line of fully autonomous cars,” said Pratt, a former MIT professor and program manager at DARPA.
Although the industry has made incredible strides over the last five years, Pratt said the biggest hurdles remain in front of researchers.
“Tomorrow, when you look across the displays at CES and you see what is currently being tested and developed, what you will actually find is that these systems can only handle certain speed ranges, certain weather conditions, certain street complexity or certain traffic. So, despite the tremendously wonderful progress that you will see, most of what has been collectively accomplished by all of us working in this field has actually been relatively easy. And the reason that it is easy is that most driving is easy. So, where we need help in autonomy and where we really need autonomy to help us the most is actually not when driving is easy. We know how to drive when driving is easy. We can do it when we are half paying attention… We need to not only solve driving when it is easy, we need to solve driving when it is hard, when it is difficult.”
For example, what happens when debris falls off the back of a truck and the debris breaks into multiple parts. How does the car respond?
“These are really hard challenges,” he said.
Up until now, Pratt said that they have measured autonomous cars based on millions of miles traveled, but in reality it is only when self-driving cars operate without incident over trillions of miles — including those bizarre unexpected incidents — that they will be ready.
It is a big challenge, and Pratt said they expect “machines to do much better” and be ever-ready and operate perfectly.
After all, there’s a lot at stake. About 1.2 million people die in auto accidents worldwide each year.
The primary goal for Pratt’s team: “making a car that is incapable of having a crash.”
To help achieve this high-level mission, Toyota’s Pratt announced four mandates as it relates to autonomous driving cars.
- Developing a car incapable of causing a crash, regardless of conditions.
- Increase access to cars to those who otherwise can’t drive, including the young and elderly
- Translate products for indoor mobility.
- Push for additional scientific discoveries in machine learning, including areas of machine science.
Toyota has partnered with MIT and Stanford as part of its $1 billion push, and at Tuesday’s press conference the company announced a slate of high-powered advisors and researchers who have joined the team. About 30 projects are ongoing with those researchers, operating under names like “Uncertainty on Uncertainty” and “The Car That Can Explain.”