The closest we have to a self-driving car on the road today is the Tesla. Models come with Auto Pilot, and a beta FSD (full self-driving) mode, which Elon Musk pushes as a fully autonomous mode. In reality, it’s little more than a fancy cruise control. And now, Tesla has admitted as much, saying Musk was “extrapolating” the autonomous abilities of the cars. Will self-driving cars ever be good enough for cities? And do we even want or need them? “The biggest barrier to autonomous vehicles in cities is their inability to navigate without human input when faced with complicated traffic patterns or unexpected circumstances,” the Electric Ride Lab’s Ibrahim Mawri told Lifewire via email. “As such, we’re not likely going to see self-driving cars become ubiquitous any time soon.”
Cities and Cars Don’t Mix
For a self-driving car to be safe, it has to know the rules of the road, know exactly where the road is, and be able to spot other cars on the road. In the city, this is complicated by the presence of people—pedestrians, cyclists, delivery drivers, kids chasing a lost ball like they’re in a 1950s public-safety film, etc. “When you consider the infrastructure for fully automated cars and trucks, you will realize that the major hindrance to the technology doesn’t lie internally within those vehicles, but rather with the environment [in which] they operate,” Ravi Maharaj, an IT specialist in the Parliament of Trinidad and Tobago, told Lifewire via email. For you and me, spotting a road is easy. But for computers, it’s a hugely complex task. They combine highly accurate maps with cameras that look at the road ahead. The computer must then work out, on the fly, what it’s seeing. It can use LiDAR to make a 3D map of the space, and artificial intelligence to better guess at what it sees, but it’s a big job. “No matter how sophisticated driverless technology becomes, with either cameras, sensors and AI navigation, they would always encounter the same issues as regular human drivers if/when dealing with the current road conditions,” says Maharaj.
No Hope
Here’s an example of how far we are from Level 5 (L5) autonomy, aka KITT from Knight Rider levels of self-driving autonomy. Elon Musk’s Boring Company has constructed a $53 million network of tunnels under the Las Vegas Convention Center, to transport visitors around the huge complex. The Teslas inside carry passengers through tunnels, which have been tailor-made for the cars, and yet they still need human drivers. It’s hard to imagine an environment better suited to a self-driving car than a controlled set of human-free tunnels, and yet Musk’s own cars can’t handle them. Essentially, the project is just Uber in tunnels. And what problem do self-driving cars solve, exactly? You can play sudoku on your morning commute, but you can do that on the bus or metro. And we already have cars you don’t have to drive: taxis. And unlike autonomous Teslas, you don’t have to own one.
Where Can Self-Driving Work?
Self-driving in cities might be doomed anyway, because cities are finally waking up to the fact that cars have no place in them. But autonomous vehicles have other uses. One is trucks. Highways are a much less chaotic environment than cities, and trucks can even drive in formation to save fuel. But the obvious case is public transit. Many public transit systems already are autonomous. London’s Docklands Light Railway opened in 1987, and operates without drivers. Many airport transit systems also work autonomously. Inter-city trains move too fast, and take so long to stop that they still need drivers, but in cities, municipal railway systems and underground metros are already mostly automatic. Drivers are there partly because they always have been, and partly because passengers feel safer with a human at the front, even if they’re not doing any driving. Perhaps Tesla will one day apply its tech to public transit, then it might be able to do some good.
