The Self-Driving Revolution — What’s up with Autonomous Vehicles?
This blog post is an adaptation of our YouTube video (linked above). Subscribe to our channel for more Lightbulb Moments like these.
We’ve been hearing about self-driving cars for some time now, but even after years of development and hype, they still aren’t everywhere, as we had imagined they might be.
Although self-driving cars may not be everywhere yet, they are much closer than many realize.
Researchers predict that we will see around 8 million autonomous cars on the streets by 2025.
In fact, the technology already exists — that is, we already have “mostly” self-driving cars on the roads, and companies like Google and Tesla have proven that it works over billions of miles of driving.
There are 6 levels of autonomous driving, with 0 being fully manually controlled, and 5 being fully self-driving, with no human intervention required. Tesla cars are currently considered level 2 self-driving because they require a human at the wheel. Google’s Waymo is considered level 4 autonomous cars, since it’s fully automated, but geofenced to specific cities only.
Tesla’s CEO Elon Musk has stated that he expects them to have level 5 autonomy within 2020.
That’s not too surprising given that Tesla has tens of billions of miles of driving data to power its machine learning algorithms.
Let’s first understand why anyone would want self-driving cars in the first place? Aren’t we getting around fine already?
A typical car spends around 95% of its life parked.
Now, imagine a world where your car can drive itself. You no longer have to be engaged in the drudgery of driving — instead, you can relax in your car, and catch up on smashing the like button for your favorite YouTubers while on your way to work. After dropping you off, your car can just go and operate as a taxi service, earning you money while you work. In fact, most people won’t even need to own a car, because everyone’s cars are now taxis. Taxi, Trucking, Moving, Delivery, and other transportation companies will save billions of dollars from not having to pay drivers.
Today we’re going to explore why we think self-driving cars are coming soon, when we expect to see them, and what the impacts of this are going to be.
The biggest challenge for self-driving cars is the extremely high stakes. Since people’s lives are actually at stake, there is an incredibly high safety expectation of these vehicles. No one wants a 4000 pound hunk of metal and glass driving around by itself unless we can be certain that the probability of error is very low.
There are around 1.35 million deaths per year worldwide due to car accidents, 94 percent of which are due to human error. So it seems reasonable to assume that a lot of those can be eliminated by highly trained computers.
Although this makes sense in theory, it takes a long time to prove. Human driving deaths are somewhere around about 1 death per 100 million miles of driving. This means it is hard to be certain that a self-driving vehicle is safe, till they’ve driven many billions of miles. But from the data we have right now, it seems pretty clear that they are significantly safer than human drivers.
Most self-driving vehicles use a combination of cameras and sensors such as ultrasonic sensors and LIDAR to get a 360-degree view of the car’s position relative to everything around it.
Tesla’s self-driving system, for example, can process 2,300 images per second, which allows for far quicker reaction times than humans are able to achieve.
Artificial intelligence algorithms are trained by analyzing billions of these situations to figure out the best move for the car, giving them more “experience” than every human driver in history. It also helps that cars don’t drink and drive.
But while they are likely safer than human driven cars, this does not mean the technology is perfect. In 2018, one of Uber’s self-driving cars in Arizona had an accident where it killed a pedestrian because it was unable to classify the object and respond appropriately. Accidents like this are rare, but they do happen. They are also a serious hurdle to autonomous driving as they undermine people’s trust, and lead to slower regulation and policy changes.
There are also some even harder moral questions that may not have clear answers. For example, consider the autonomous car’s version of the trolley problem:
If there is someone in the car’s way, and the only way to save that person is to swerve left and risk the passenger’s life, what should the car be programmed to do?
What if the only way to save a pedestrian is to swerve and risk another pedestrian’s life. These are moral problems that people have been debating for centuries.
For technologies like these, the catalyst for adoption is often business need. Companies such as Uber and Lyft, for example, are banking on the future of autonomous vehicles to achieve profitability. Tesla also intends to start its own competitor to Uber in the near future with a self-driving taxi network.
Trucking is another industry that stands to benefit greatly from self-driving. The trucking industry makes nearly a TRILLION dollars in revenue annually.
Truck drivers drive around 2000–3000 miles a week on average, getting paid around 30–40 cents per mile they drive. That’s around $800 per week. America has around 3.5 million truck drivers.
This means that the trucking industry could save somewhere around $2.5 BILLION dollars per WEEK or over 100 billion dollars in a year.
Does this explain why Tesla announced that they are launching a semi-truck and a pickup truck soon?
Eventually, the world will always gravitate towards efficiency. And our current patterns of owning cars are neither efficient, nor sustainable. We have far too many cars in the world, most of which spend most of their time parked.
Studies have estimated that there are somewhere in the neighborhood of 1 BILLION parking spaces in the US alone. That’s tens or hundreds of billions of square feet of real estate. Very inefficient.
Self-driving cars as a concept have been around for decades. And it’s a no-brainer if it works well. Recent waves of innovation in image recognition, artificial intelligence, and GPUs to power machine learning have brought autonomous cars into the real world. The two major players in the space right now are Tesla, and Google, and both have different models. Tesla sells cars directly to consumers and then gathers data for it’s AI while they drive. Google’s Waymo has been operating a completely driverless taxi service in certain cities for a while now, and they’re using that to research and test their technology.
Who do you think will win that race? Would you feel comfortable being driven around in a self-driving car? Let us know in the comments below, and be sure to subscribe to our YouTube channel for more Lightbulb Moments like this. Stay curious, and we’ll be back with more soon.
