I rode in a self-driving car for the first time, and it was operated by the Google of Russia - here's what it was like
This is what the car, a converted Toyota Prius, looked like.
It had three lidar sensors, which emit pulses of light to detect objects and determine how far they are from the car.
The car had five cameras, like the one pictured below.
It had a computer in the trunk.
Inside, there were two iPads that illustrated part of what the vehicle was seeing.
The ride was uneventful, and the route generally avoided difficult maneuvers like unprotected left turns and merges into busy lanes.
Had I not made a point of observing the car's performance, I likely would have forgotten that it was driving without human assistance. There was a safety driver in the driver's seat, but I did not see his hands did not touch the wheel at any point during our ride, and while I couldn’t see his feet, I had no reason to believe he was controlling the car’s braking and acceleration.
The car felt like it was being driven by a cautious human.
The car often drove slower than surrounding traffic (it briefly reached a top speed of around 45 mph), kept a safe distance behind the car ahead of it, and stayed within lane markings.
Aside from a few slightly abrupt stops, the ride didn’t have any major hiccups, and it suggested that Yandex’s technology could, at the least, be ready to carry passengers on short routes that won't present many challenges.
Of course, I would have to observe more than one ride to know that for sure, but limited-scale, autonomous ride-hailing and shuttle services from Waymo and startups like May Mobility and Optimus Ride have indicated that current autonomous driving technology can handle relatively small, geofenced areas.
Dmitry Polishchuk, Yandex's head of self-driving, said the company had not yet determined if it will keep testing in Las Vegas or elsewhere in the US after CES.
The company is currently running two autonomous ride-hailing services in the Russian cities of Innopolis and Skolkovo (the former covers 0.8 square miles, and the latter covers 2.38 square miles). Yandex hopes to expand its ride-hailing services to Moscow and Kazan when it receives regulatory approval, which could happen in 2019, Polishchuk said.
He also said Yandex vehicles may be able to drive as well as a human in certain cities within four years.
Challenges remain for Yandex's autonomous driving division, the two most difficult of which are developing a vehicle's ability to make decisions and predict what vehicles and people around it will do.
"This is where all the competition will happen in the future," Polishchuk said. "This competition in self-driving is not about hardware. It's about software."
But Yandex has made a significant amount of progress in just a few years.
The company began working on autonomous driving technology in January 2017, Polishchuk said (there appears to be some internal disagreement over whether the work began in December 2016 or January 2017, based on conflicting press releases, but Polishchuk insisted the company's autonomous driving program began in January 2017), and by August 2018, the company launched what it says was the first autonomous ride-hailing service in Europe. Yandex has since given over 2,000 rides in autonomous vehicles.
Polishchuk said Yandex was able to start giving rides to customers so quickly because it had previously developed products that correspond to some of the capabilities needed for self-driving cars, including computer vision for image search, a navigation app, and a localization feature that can determine where people are using its search engine.
While Polishchuk said Waymo is the current leader in autonomous driving technology, he said he believes Yandex can compete with Waymo's autonomous driving software.
"We are competing with Google for [search traffic in Russia] all the time, and we are beating them, so we know how to do this stuff," he said. "We are happy to compete."
Next Story
