Self-driving cars have been on our minds for a while now, and manufacturers seem to be making progress every day. But when will you have a self-driving car in your driveway?
Well, the answer is a bit complicated. To answer it, we need to understand where we are with self-driving cars right now, and what steps we need to take to make fully autonomous cars a reality in the future.
When is a Car Fully Autonomous?
According to SAE International standards, autonomous cars are rated with a “level” on a scale from 0 to 5. Level 0 cars have no autonomy and must be controlled by a human at all times. Level 5 cars are fully autonomous and don’t require any help from a human to operate.
The car that you’re driving right now is, more than likely, a Level 1 autonomous vehicle. It has cruise control features, and it might even have a backup camera. But there’s a chance that you’re driving a Level 2 or Level 3 semi-autonomous car, like a Tesla, a Cadillac CT6, a Mercedes-Benz E-Class, or a Volvo S90. These cars have features like Auto Pilot or Super Cruise, which allow you to take your hands off the wheel while your car maintains a steady speed and changes lanes.
Okay, glorified cruise control is super cool, but you’re trying to buy a fully self-driving car. Are there any Level 4 or Level 5 autonomous cars on the market? The answer is a resounding “no,” but you could find yourself in a genuine self-driving car very soon.
2019 is the Year of Level 4 Autonomous Cars
We haven’t talked about Level 4 autonomous vehicles yet, but they’re an important step in the journey to self driving cars. The dividing line between Level 4 and Level 5 autonomous cars is, essentially, flexibility. While Level 5 autonomous cars can traverse anywhere on a whim and safely maneuver through novel locations, Level 4 cars are stuck in predetermined areas, like a city or a state. Level 4 cars are also fed roadmaps and travel data that make everything super easy and safe. While Level 5 cars are comparable to hyper-aware human drivers, Level 4 cars are comparable to a near-blind person walking the streets of a familiar city.
That’s why Level 4 autonomous cars are being built specifically for driverless ride-sharing and delivery services, not for the retail car market. These cars run on a predetermined track, but they’re capable of maneuvering between lanes and avoiding collisions with cars or pedestrians. They’re reliable, safe, and good for collecting data that will be useful for Level 5 cars in the future. Manufacturers are already using Level 4 autonomous cars in some cities, and there’s a decent chance that you’ve seen (or been inside of) a Level 4 autonomous car.
Waymo, the self-driving car company that was founded by Google, is one of the forerunners of Level 4 self-driving cars. The company has been working on autonomous cars for almost a decade, and Waymo vehicles are currently used for ride sharing (by a select group of people) in Chandler, Arizona and San Francisco, California. Residents of the Metro Phoenix area that want free rides in a Waymo autonomous taxi can sign up for an Early Rider program right now, but spaces are limited, so don’t spend too much time thinking.
Waymo’s biggest competitor, GM, is releasing a fleet of self-driving Chevy Bolts this year. These Chevy Bolts have no steering wheels, pedals, or speedometers, and they mark a confident step toward driverless cars. We don’t know where GM is going to test these cars, but the company has been using Level 4 autonomous cars (with steering wheels) in San Francisco and Phoenix for a few years, so there’s a chance that the newest Chevy Bolt will end up in California or Arizona.
But if you aren’t a resident of Arizona or California, then you may see a different brand of Level 4 autonomous vehicles in your area. Right now, Uber focusing on a fleet of self-driving cars in Pittsburg, as an accident in Arizona forced the company to downscale testing a bit. Ford is teaming up with Walmart to deliver groceries from a Fusion Hybrid Sedan in Miami, and there should be Ford autonomous taxis driving around by 2021.
There are a few other companies that plan to bring Level 4 autonomous cars to American streets in the next few years, including Nissan and Volkswagon. But they’re a little late to the game, and they haven’t announced which cities they plan to test their cars in.
Where Are the Level 5 Autonomous Cars?
The number one reason why you don’t have a Level 5 self-driving car in your garage right now is simply because they don’t exist yet. According to SAE standards, a Level 5 autonomous car needs to perform like a human in novel locations, and it needs to be able to respond to wild hazards as a human does.
But let’s be realistic here. Level 5 autonomous cars need to outperform humans at every turn, stop, and wild act of God before lawmakers will allow them on our streets. We need data that shows how a car controlled by a human is categorically more dangerous than a car controlled by a computer.
So, what needs to be done? Well, for starters, manufacturers could send their safety reports to the US Department of Transportation. Until the DOT has a good collection of safety data, laws won’t get passed. Understandably, not every company has enough data to submit annual reports to the DOT, and only Waymo and GM have managed to step up to the plate.
But the biggest hurdle for autonomous car manufactures isn’t the DOT, it’s AI. Naturally, AI will get better over time. Here’s the catch, it needs to get better at thinking like a human, and developers need to get better at thinking like AI.
We’ve built or roads and highways for the use of humans, not robots. As a result, our understanding of the “easy” and “hard” parts of driving are totally subjective. While manufacturers are focusing on how AI can make rides as smooth as possible by adjusting tire suspension in real time (which is pretty cool), they’re ignoring the fact that AI may be completely unreliable in some average, real world driving situations.
For example, remember the Tesla that crashed into a truck on the highway while in Auto Pilot mode? That happened because the Tesla couldn’t distinguish the shiny side of a truck from the sky. But Tesla’s don’t just use cameras to get around, they also use radar (unlike competitors, Tesla doesn’t use LiDAR). Before this wreck happened, the Tesla’s radar systems detected an obstacle, which should have prompted the brakes to activate. But Tesla designed their radar system to ignore large horizontal objects, as most of the large horizontal objects that you see on the highway are simply traffic signs. So while Tesla took the time to fix an obvious problem, they ignored how their solution might complicate the AI’s ability to make good choices.
Manufacturers also have to teach the subtle language of the road to AI. You communicate with other drivers all the time, whether you realize it or not. When another driver flashes their blinkers at a four way stop, they’re communicating that you have the right of way. When you go to merge on the highway, other drivers won’t let you in unless you use a bit of confidence and aggression. If an autonomous car doesn’t know how to use subtle forms of communication on the road, then it will never be reliable or safe. Of course, the best solution to this problem is to replace every car on the road with a self driving car, but that won’t happen for decades (at least).
Although you can’t own a self-driving car right now, you can enjoy a taste of the future in a Level 2 or a Level 4 autonomous vehicle. Those opportunities aren’t available to everyone just yet, so if you’re a less-than-lucky autonomous car enthusiast, then you’ll have to settle for photos from the Detroit Auto Show or CES 2019.