We Automate the Driver, Not the Car

The purpose of innovation is to make our lives safer, easier, and more convenient. If something is dangerous or just plain boring, you can bet an engineer somewhere is working on a way to automate it. Generating a proof of concept and implementing that solution are two different processes, however. Sometimes we get so caught up in the technology itself we forget there’s a goal waiting on the other side.

The automated vehicle industry generates a ton of hype. Every few weeks a new article talks about how self-driving cars are right around the corner. The reality is the technology needs more time to mature before real automation can take over. We’re not automating cars, after all, we’re automating drivers. This is vastly more complicated than adding automatic windows or tacking on cruise control features.

Driver’s emotion often helps cope with hard road situations. Can AVs solve all edge cases reliably with their one-size-fits-all algorithms?

Between now and the inevitable self-driving future, we’re going to have to figure out how to mimic a human driver’s capabilities using sensors and an on-board computer. As you can probably guess, that’s no small task.

Automating via Technology

The first passenger elevators were operated entirely by humans. Someone actually had to stand inside and pull open the doors to let people in, manually direct the cab to the correct floor, then open the doors upon arrival. It was an arduous process by today’s standards, and indeed it became obsolete only a few decades later. What’s surprising is that when automatic elevators were first installed, people hesitated to use them. How in the world can you replace an operator with nothing but wires and buttons?!

Can replacing the driver in autonomous vehicles be as easy as replacing the elevator man with buttons?

A century later we’re asking the same questions about more complicated endeavors. Virtually every modern auto maker is working on some version of a self-driving vehicle, whether it be a passenger car, smart taxi, or cargo transport. The National Highway Traffic Safety Administration has even laid out five stages of development self-driving cars will go through before they’re fully automated. We’re somewhere between steps two and three right now. Features like electronic stability control are common, while driver assistance through automatic braking and lane keeping are gaining in popularity. Step four, full automation, is a big one, and it’s not exactly right around the corner, either.

Some vehicles operate with limited automation under good driving conditions. This arguably fuels most of the hype surrounding self-driving vehicles. A car being able to drive on a sunny freeway in low traffic is a far cry from one that can take you through the countryside in the snow. Ultimately, we need autonomous vehicles that can take us anywhere, any time, in any conditions, and on any road, just like a human driver.

Getting to the Goal

To tackle full automation we need to focus on replacing the driver, not the car itself. Engineers have identified three main areas that must be addressed on our way to full automation: navigation, control, and hazard avoidance. Human drivers do all of the above with our eyes, our brain, and our reflexes. Naturally, we are creating sensors and software that initially aims to replicate these features as efficiently as possible.

Let’s take a quick look at the main areas of self-driving technology and their current status, starting with artificial intelligence. We’re nowhere near the technological singularity sci-fi novels have warned us about. Fortunately, we don’t need full artificial intelligence to drive our cars, just smart software that can coordinate and process data fed to it by the vehicle’s sensors. The human brain far outperforms modern machine in its ability to evaluate complex navigational hazards on the road. AI might be faster, but its accuracy and adaptability need serious improvements before it’s ready to roll full-time.

Lidar, radar, cameras, and related sensors on autonomous vehicles aim to replace a driver’s vision. Surprisingly, these do a good job of providing data to the car’s main computer, provided the roads themselves are cooperative. Snow, rain, and non-standard road features are still a problem for much of this technology. A human driver, by contrast, can travel through all of this and more with sight alone. Our technology is getting there, but accurate enough sensors combined with the right evaluative mechanisms still lag behind human performance by a considerable margin.

Path to Automation

Autonomous vehicles represent a massive shift in mobility, one that we are only now beginning to understand. Self-driving cars could change how we think about vehicles, even alter the shape of our cities. But before that happens, we must realize what this technology is trying to do — automate a driver’s senses, reaction times, and navigational ability — and just how monumental of a job that is.

Replacing humans in the driver’s seat is a far more complex task than replacing an elevator operator with a panel of buttons. We absolutely should be excited about the advancements, especially with the leaps we’ve made in the past few years and what they could mean for the future. Tempering expectations is crucial, though, as the path to full automation is a gradual one, and we still have a great deal of ground to cover.

Ultimately, we will have to replace the driver with autonomous vehicles. But how long is the path ahead?

Once we do automate the driver and place that technology in our vehicles, there will be nothing stopping us from going further. Sensors and computers don’t suffer the same limits as our biology. Fully automated trips in electric vehicles through the outback are not out of the question. But for right now, let’s focus on the task at hand. The future of human mobility is evolving, and we’ve got front row seats for the show.