Article

Understanding Autonomous Vehicle Levels

Ben Parker

The Society of Automotive Engineers (or the SAE) set the 6 levels of driving automation ranging from no automation to fully autonomous vehicles. The United States Department of Transportation has now adopted these standards as well.

Credit Ben Parker

Article Highlights

There’s a future that you and I will most likely see where there won’t be a driver in the Uber taking you to the airport, or home from the grocery store. The implications of self-driving cars are huge; from environmental, to safety. Just like how my iPhone is better at remembering things than I am, a self-driving car is better at remembering to pick up the kids from soccer practice. What a strange, exciting world that will be. Until then, there’s a range of autonomy when it comes to self-driving vehicles. I spoke briefly about this in my article “How Self Driving Cars Work” as well.

Level Zero - The Toyota Tercel You Drove To School

This level is probably most of the cars you’ve driven in your lifetime. Like the infamous Toyota Tercel, my best friend drove all 8 of us guys around in high school. In level zero you are in full control of the steering, brakes, and acceleration. 


Level One - Cars Made In The Past Three Years

Most cars made from around 2017 usually have some features of level one. This includes lane-assist, brake-assist, or adaptive cruise control. The vehicle has the ability to control steering, braking, and acceleration, but no two at the same time.


Level Two - Tesla Model S

This is the one you’re probably most aware of out there. This is the level that people are getting a little too comfortable with. If you own a Tesla (I do not) you know what I mean. This level requires the full attention of the driver so you could take control at any time, but the car is able to accelerate, brake, and steer on its own. 


Level Three - Nothing You’ve Ridden In

This is a car that it’s legal for you to take your eyes off the road while driving -- or not driving I guess. This combines all the features of the previous level, however, the computer is more capable and able to fully navigate itself with a driver. 


Level Four: High Automation

This is the closest we’ve come to that future of your minivan going off to go grocery shopping without you. Level four cars can do all the driving, but as of today, these cars can only do all the driving in the right conditions, good weather and good roads. This is what you’re hearing/seeing about in the news the most. This is what Waymo (Google), Uber, Lyft have all produced or prototyped and are testing. Again this is only level four, not five because of the conditions of good weather and good roads. 


Level Five: Doesn’t Even Have a Steering Wheel

This is the future. This is the dream. The car doesn’t even have a steering wheel. The car can fully operate on its own. At this point, I wonder if we should even call it a car it’s really more of a robot than a car.


Prometheus, or “Programme for a European traffic of highest efficiency and unprecedented safety” was perhaps the first dream of the self-driving car. Daimler-Benz AG created the project that developed a lot of technology that is used today.

Credit Daimler

“October 1994 when the research vehicle covered more than 1,000 kilometres on a three-lane motorway in normal traffic at speeds of up to 130 km/h while demonstrating lane changes in both directions as well as autonomous overtaking after approval by the safety driver.”

Credit Daimler

Some speculators argue that these Levels of Automation perpetuate the idea that these cars are truly ready to drive themselves and that responsibility should be placed on the car over the driver.

The Potential Problems with the Levels of Automation


There hasn’t been a lot, but there have been a few fatal accidents due to people relying too much on their cars’ self-driving features. Some speculators argue that these Levels of Automation perpetuate the idea that these cars are truly ready to drive themselves and that responsibility should be placed on the car over the driver. One paper was written by Marc Canellas and Rachel Hagg from Communications of the ACM explaining some of these issues. In their paper, they write, “The central flaw of LOA is right there in its name. Levels of automation focus on a singular static definition of the automation capabilities, ignoring the deeper ideas of teamwork, collaboration and interdependency necessary [to safely operate]”. 


Thomas B. Sheridan (who created the Levels of Automation) has spoken out to explain his intention was never that the levels of automation would be a framework by which automation should be constructed, or measured in this context. The basic issue is that the levels of automation work as sort of a means by which responsibility is spread vaguely between the driver and the car. The driver and the car should really equal parts of a team. One example could be that you’re running a competitive relay with someone, they tell you they’ve got it most of the way, you should just sit back and if you’re really needed they’ll call on you, but most likely you won’t be needed. The last 100 meters of the race is really close between all competitors, but instead of a typical heads up from your teammate, he just suddenly drops out and you’ve got to get up to an all-out sprint in seconds to catch up. This is basically how partial autonomous vehicles work. 


In the second level of autonomy, it dictates that the car will drive itself and the driver should “at all times” monitor the rest of the responsibilities. This oversimplifies the remaining responsibilities that a human driver is currently responsible for. As humans, we’re not that good at monitoring our surroundings for extended periods of time. The car is designed to assume that the human will monitor the surroundings better than if not as good as a computer could itself. None of us have attention spans long enough.

article continues below Advertisement

Voyage Auto has successfully used their driverless systems to operate within two senior communities in California and Florida.

Credit Voyage Auto

This means that you’re going to be paying for Spotify $10/month and then your car subscription, and then all the other subscriptions you pay for as well.

Will We Own Fully Autonomous Cars?

I think this is still up in the air. From where it looks right now it seems that the autonomous vehicle makers are bearing towards becoming services companies over traditional automotive companies that sell cars. This means that you’re going to be paying for Spotify $10/month and then your car subscription, and then all the other subscriptions you pay for as well. There are some real pros and cons to this. These are only a few of many that are mentioned in my other article here.


Pros:

  • Auto Insurance: There will no longer be a need for us to pay for car insurance. As Uber, Lyft, and other shared-economy companies do currently, these companies will need to absorb those liabilities. This will be an interesting shift for those insurance companies as many may shift to insuring the manufacturers themselves.


  • Car Upkeep: Most people pay roughly $817/year to maintain their cars currently. 

  • Time: Imagine being able to have 40 minutes extra a day to do whatever you want. When you’re not the one driving you’ll be able to work, relax, read, whatever in the back of the self-driving Uber. NHTSA reported that Americans spent almost 7 billion hours in traffic in 2014, and I’m sure that statistic hasn’t gone down. Mckinsey Consulting did a study that reported autonomous vehicles could remove somewhere close to 50 minutes of traffic a day. 


  • Home Design: This one could also be a con depending on who you ask. What will homes look like when we don’t need home garages? 


  • Mobility: This should be at the top of the list but providing the ability for many who are physically or mentally impaired the ability to drive and get around. 


  • Safety: This is another huge consideration, 94% of all serious motor vehicle crashes are due to human error. Humans can only see roughly 135 degrees horizontally and 180 degrees vertically, when an autonomous vehicle has a 360-degree image of its surroundings.


  • Parking: Another benefit brought up by Mckinsey is freed parking space. This goes hand in hand with no more home garage. They estimate that billions of square meters of parking will be reduced. 


Cons: 

  • Subscriptions: We’re already living in a time of subscription fatigue. I can’t imagine having to pay what I imagine will be over one hundred dollars a month just for the ability to drive to work in the morning. 


  • Availability: A big question is how will this scale, how can all of us be needing and using autonomous vehicles. Will varying schedules and destinations mean more cars on the road?


  • Time: In tandem with availability, you may have to wait for your autonomous car to come to pick you up for work while it’s chartering someone else too and from Equinox. 


  • Security: A weird dystopian reality will be security. Virtually anything that is a computer is susceptible to hacking. I don’t imagine this being a huge issue, but one worth noting is that a computer is perhaps to some degree more easily manipulated than a human. 


This list is some of many of the pros and cons that we’re considering moving into a more autonomous world. But many more questions are still left to be answered. Technology in America seems to move like lightning when policy moves like sludge.

Olli is a driverless shuttle that is geared towards operating on campuses for schools, businesses and hospitals.

Credit Local Motors