How long until we have roadworthy autonomous cars?
When you hear about self-driving cars, you might imagine something completely autonomous — just tell it where you want to go and you can spend the drive drafting an email, watching a movie, or just planning your day. The National Highway Traffic Safety Administration, however, divides self-driving cars into five distinct levels. And despite a few examples of the autopilot pulling off some amazing moves, a completely self-aware car that can handle everything with no human input is the fifth and final level – and it’s still a long way off.
Some of the technologies required to realize this level of autonomy are already available in cars today, but many have a long way to go before they can be deployed to vehicles on public roads. Here are a few of the crucial technologies needed for level-5 self-driving cars to become a reality.
Communication Between Cars
The safety features that make self-driving cars attractive come largely from the potential for network-aware vehicles to communicate their next move ahead of time, and to monitor nearby cars for erratic behavior using lasers, sensors, and other technologies. More and more engineers are developing materials to better protect both car sensors and passengers, which in turn increases the likelihood that autonomous cars will hit the roads soon.
By removing the potential for human error that comes with our need to react to cars in other lanes, network-aware cars should greatly lower the likelihood of an accident. Network technology in cars has improved in recent years, but it’s nowhere near where it needs to be to solve this problem. The easy answer seems to be a large municipal network. However, it’s still uncertain who would build such a network and how the operator of that network would recoup its costs. It also means that fully autonomous cars would immediately run into issues when they were outside of the range of that municipal network.
Connected highways might solve some of that – and could make sense if and when automated commercial semis are introduced – but it would again run into funding problems. A private entity might find a way to make it profitable, but otherwise, it would require massive federal funding to upgrade the nation’s infrastructure.
One of the biggest and most well-documented challenges in making self-driving cars roadworthy is defending them from network threats. While this is related to the question of how to facilitate car-to-car communication, it’s a completely daunting task in and of itself.
A well-planned cyber-attack on even a single car could potentially cause a highway-closing accident. Combine this with the need for a system to quickly patch cars against threats while they’re on the road, and for authorities to disable any car at will should the car’s autonomous systems fail, and you have a multifaceted can of worms for the self-driving car industry.
The network that connects the cars together could also be vulnerable to an attack, and if that happens it could affect millions. It’s still all theoretical, but if the near-constant cyber attacks on everything from news agencies to cryptocurrency to political parties, something as simple as a DDoS attack could cripple an entire city or more. In Q1 2017 alone, DDoS attacks have risen 380-percent, and there’s no reason to think that will decrease as we become even more connected.
Researchers have investigated the issue of cars recognizing other cars on the road, but what about pedestrians? People in the crosswalk can’t be expected to communicate their next move to an oncoming car wirelessly, and autonomous cars are being programmed to do everything possible to avoid hitting them.
Google and other pioneers in the self-driving car industry are working to design image-recognition systems that will allow cars to pick out the shape of a human and avoid pedestrians even in the dark. Not only could this system save lives, it will assist cars with reading signs and identifying road markings to help them track accurately on the road.
Two leading technologies that are being touted as solutions to this issue are LIDAR — a laser-based solution — and camera vision. There is a downside, however. In crowded cities where pedestrians frequently run into the road – New York City as a moderate example, Tokyo and Shanghai as extremes – this could bring autonmous cars to a grinding halt.
What should a self-driving car cost? This is a difficult question because the nature of vehicles changes when you remove the human interaction at the wheel. It also depends on the quality of the car and the pedigree. A BMW with an autonomous package will almost certainly be far more expensive than a Chevy sedan, for example. There may also be a price hike for larger cars like an SUV that may require more sensors than, say, a coupe. And of course, there’s the premium people will pay to be early adopters.
Initially, it’s likely that cars equipped with level-4 and -5 systems will be quite expensive, with prices being driven up by the sheer cost of the technology. There may also be a monthly fee to keep the car connected, and until there is a decent amount of history to go on that proves the safety of autonomous cars, the insurance may be pricey.
How Do I Drive the Self-Driving Car?
You’ve probably assumed that getting somewhere in one of these futuristic cars is as simple as telling the navigation system where you want to go, but we can’t yet say for certain if that’s all. There are aspects of operating these new cars we haven’t thought of yet, and their success in the market depends on manufacturers’ ability to make them simple and intuitive to operate.
Perhaps the biggest question is how long will it take for drivers and passengers to accept and completely trust their autonomous cars. It’s one thing for a car to take control with a test subject behind the wheel nervously waiting to see if they need to grab the wheel, it’s another to trust it so deeply that you are willing to completely turn your attention away from the road and onto something else. There’s no shortcut for that.