The self-driving car club is getting bigger everyday.
For a futuristic technology that hasn’t really hit prime time, there are a lot of companies involved in autonomous driving. CES 2019 was littered with players – from small startups to large automakers – touting autonomous cars, components, or possible uses. But while most of those players close down their booths and get out of town at the end of the show, Lyft and Aptiv get back to work.
Lyft and Aptiv brought self-driving cars to Las Vegas during CES 2018 – and never left. They began giving rides to the public in May 2018, allowing anyone with the Lyft app to hail an autonomous car. Since then, Lyft and Aptiv have given thousands of rides, and demonstrated how self-driving cars could work in the real world.
“We are together the largest commercial self-driving car operation today on public roads,” Jody Kelman, director of Lyft’s self-driving platform, told Digital Trends. But while the program has been successful, it also shows just how far the technology still has to go.
Deploying autonomous cars is no small undertaking. The technology itself is very complex, and getting it on the road involves an equally complex balancing act of government relations, fleet logistics, and customer service. That’s why Lyft and Aptiv decided to team up.
Aptiv operates the cars and developed the systems that allow them to drive themselves. Under its previous guise as the automotive supplier Delphi, the company completed a successful coast-to-coast drive between San Francisco and New York, with a car in autonomous mode a claimed 99 percent of the time. In December 2017, the company rebranded as Aptiv to focus more on tech, spinning off its more traditional automotive parts business.
Lyft and other companies view autonomous driving and ridesharing as a natural combination.
Meanwhile, Lyft arrived on the scene as a competitor to Uber in ridesharing, and became very interested in self-driving cars. Lyft and other companies view autonomous driving and ridesharing as a natural combination. Operators don’t have to pay human drivers, and cars can stay on the road earning money longer. At the same time, companies maintain more control over how cars are used, a crucial consideration in the rollout of a new technology. That’s why you’ll likely hail a self-driving car long before you can buy one – if the technology achieves large-scale commercialization at all.
Lyft has its own in-house autonomous-driving program, but the company likes working with others. It has signed agreements with Waymo and automotive supplier Magna. It also partnered with startup NuTonomy on a Boston autonomous ridesharing pilot (Aptiv predecessor Delphi bought NuTonomy in October 2017).
That’s how we ended up walking up to a black BMW 540i on a cloudy January day in Vegas. Aptiv currently has 75 cars like this in the city, with 30 operating in Lyft service.
At first glance, only the car’s brightly colored wheels, Aptiv decals, and the red license plates that denote all self-driving cars registered in Nevada, indicated that this was anything other than a normal luxury car. But upon closer inspection, the body is covered with sensors. They’re in the bumpers, under the mirrors, and even in the nostrils of the BMW’s grille.
“We retain the beauty of the BMW,” Jada Smith, Aptiv vice president, advanced engineering and external relations, said of the car, with more than a hint of pride. All of today’s self-driving cars are prototypes, so companies typically don’t put much effort into making them look attractive. But that’s something Aptiv believes its customers want, Smith said.
“They don’t want a tin can on top of the roof.”
Making an autonomous car look (relatively) normal is impressive, given how much stuff Aptiv adds to a vehicle already crammed full of its own electronics. Aptiv’s sensor suite includes nine lidar units (four short-range, five long-range), 10 radar units (six electronically scanning radars, four short range radars), a trifocal camera, a camera specifically tasked with reading traffic lights, two GPS antennae, and two computers in the trunk. The picture painted by these sensors (in vivid, acid-trip colors) is displayed on a screen atop the dashboard. It’s meant to build confidence in the car’s capabilities by showing how much it sees, Smith said.
If you hail one of these cars, prepare to be bored.
The car also has one antenna for dedicated short-range communications (DSRC), which allows it to “talk” to surrounding infrastructure. Thanks to DSRC, cars know whether a traffic light is red or green even if they don’t have a direct line of sight (a similar system is already available in some Audis). Tricks like that have automakers and tech companies convinced that DSRC tech, also referred to as V2V (vehicle-to-vehicle) or V2X (vehicle-to-everything) should be installed in production cars. Aptiv worked with the local government to install over 100 DSRC sensors around Vegas, but doing that all over the country would be costly, and governments in other regions could be less friendly.
Those sensors wouldn’t do much good without software to interpret their data and, well, actually drive the car. The goal is to create software that is not only competent, but also predictable.
“It’s very similar to how we learn to drive as people,” Smith explained. Engineers “teach” the software the rules of the road first, and then teach it driving techniques. “It gives us a very traceable and explainable code set,” Smith said.
What does all of that mean from the back seat? It means that if you hail one of these cars, prepare to be bored.
The novelty of a person sitting in the driver’s seat (each car has a human safety driver onboard at all times) while the wheel turns of its own accord fades quickly. The end result of this intense concentration of technology is a car that performs like a very cautious human driver.
Aptiv programs the system to be conservative, and to stick to speed limits (25 mph on most of the Vegas streets where we rode). Over the past year of operations in Vegas, the company claims to have tweaked the system for fewer sudden stops (although we still noticed some) and to give pedestrians, cyclists, and other vehicles a wider berth when passing.
“The car is more human like in the way it navigates the road,” Abe Ghabra, managing director of Aptiv’s Las Vegas operations, said of the changes.
Passengers seem to like it. As of January 2019, about 30,000 Lyft customers have ridden in Aptiv’s self-driving cars, and given the experience an average rating of 4.95 out of 5 stars, said Jody Kelman, the director of Lyft’s self-driving platform. She said one passenger had taken 14 self-driving car rides to date, while another told the company that riding in an autonomous car was the third item on his bucket list, alongside marrying his wife, and sky diving. But Lyft’s goal is to make the experience unremarkable.
“This should really be as normal as pulling out your Lyft app and requesting any ride,” Kelman said.
The end result of this intense concentration of technology is a car that performs like a very cautious human driver.
It already is, in a sense, but not for the reasons Lyft and Aptiv likely want.
Under an agreement with the various hotels and casinos along the Las Vegas Strip, cars only operate in manual mode on private property. It’s a sensible decision, given that these businesses are where most Lyft riders want to go, but it means most rides are short trips where a human driver is in control much of the time. When the car is in autonomous mode, it’s likely traveling in a straight line, at slow speeds, in stop and go traffic.
Unusual circumstances can also force human drivers to take over: on one of two demos organized by Aptiv and Lyft, this happened when a car reached a new construction zone. Aptiv simply hadn’t had time to let the system know that it was there.
This doesn’t make for the most impressive demonstration of a technology that’s supposed to change the world, but it may be good enough for now. Lyft’s Kelman said self-driving cars will comprise part of a “hybrid network.” Lyft will deploy them where they make sense, and use alternatives like conventional vehicles, bikes, or scooters to fill in the gaps. Customers care more about getting where they want to go quickly and cheaply than the specific mode of transportation, she said.
Long road ahead
Cruising down the Las Vegas Strip in a self-driving car – and having it be a nonevent – is mind-boggling. Not long ago, this technology was thought to be impossible. Now dozens of companies are involved in developing it. But autonomous cars still have a long way to go. The cars operated by Lyft and Aptiv have lots of limitations, and they don’t even have to deal with, say, the winter weather of Detroit or the take-no-prisoners attitude of New York City drivers. Transporting paying customers in self-driving cars represents a giant leap for the technology, but several more are needed for autonomous driving to really make an impact.
- Waymo becomes the first company to charge for rides in self-driving cars
- Did that car just wink at you? Daimler previews car-to-pedestrian signals
- Baby, you can’t drive my cube: All the insane self-driving lounges at CES 2019
- Waymo’s autonomous cars are coming under attack in Arizona
- U.K.’s ‘advanced’ self-driving car trials won’t require human safety drivers