Self-Driving Hubris
Humans found horses 30,000 years ago. For the most part, the horses served as food. In South America, horses were hunted to extinction. It was not until 3500 BC that horses were domesticated mostly in the Steppes. Once domesticated, horses were highly versatile and fast land animals that served as great transportation. They were very useful on plains but they also could climb hills and other terrains, just not as swiftly.
When the first automobiles were developed, nobody bothered to find out how the car would gallop over obstacles or if it would be able to drive across a not-so-shallow stream.
Imagine the over-engineering that would have been needed to create a car that could jump over a haystack and drive through chest-high waters!
Instead across the world, the most widespread infrastructure development initiative was unleashed and we got the Road Network. Today there are about 64 Million kilometres of roads across the world. America has the most at 6.8 million Kms followed by India at about 6.3 million Kms.
Going faster and being safer meant having better infrastructure, not teaching the car to gallop.
But Hubris has no ends
Lidar is a technology that has been around for decades and has been getting better year after year. It uses a laser to map the topography around its camera. You can strap on a lidar sensor to a vehicle and keep rendering the topography around the vehicle to know where the obstacles are. This is one way of “seeing” the world.
Then there is Tesla where Elon Musk wants to build computer vision systems that can “see” the world as humans would. How has it been going? Well, check out the video below.
There is a hoard of Musk fans who have trolled the hell out of the post.
I personally feel that neither of the technologies is the solution.
Talking Cars
Just like cars needed road and not the ability to gallop, autonomous vehicles will need infrastructure that can communicate with the vehicles rather than the ability to see. The roads should be able to talk to the car with embedded sensors that have information about the surroundings.
Also, vehicle-to-vehicle communication would be important since it would make it possible for a mesh of information to exist which would be far greater than what one vehicle can sense.
But we live in a world of ‘defensible moats’. Everyone wants to own proprietary technology. Imagine a proprietary internet. It would have sucked.
While the hubris-laden morons were fighting with each other over what is best, much like the tweet above, the policy changed.
Today, the DC circuit court released its decision, siding with the Federal Communications Commission, on its reallocation of part of the 5.9GHz band. It’s a big win for the FCC and a big loss for the auto industry, which has promised to use the airwaves to improve safety through a technology called “vehicle-to-vehicle” (V2V) or “vehicle-to-everything” (V2X) communication.
The problem, as hilariously put by Judge Justin Walker in his opinion, is that this technology has never really existed. It was one of those “just around the corner”-type innovations that has always been promised but never actually delivered. It was a fantasy, and today, the court’s basically said as much.
Source: Verge
The reason it was always around the corner is that such technology would require adhering to a standard and agreeing to work together. Nobody could.
At least in the US, the window of opportunity has closed.
Technology comes with its inherent limits. You need to build a world where it can thrive despite those limits and not overcome those limits.
