Brad Templeton: Changing the world with robocars
WARNING: Liveblogging – prone to error and inaccuracy. Will be updated/improved over the next 48 hours.
What happens when computers meet cars?
It turns out human beings are pretty bad drivers. 1.2m people are killed on the roads every year. In the US we spend 50bn hours driving. That's astounding. In Germany, there are 3,400 deaths on the road a year - about 65% of the American rate. 40% of road fatalities are due to drinking - robots don't do much of that.
We're losing 41 hours a year to congestion. We're using land and energy and emissions to support our current car use. Humans are driving 1.7 lights years a year.
10 years ago the US military held a contest for self-driving cars. It didn't go well. But, by the second contest five autonomous cars completed the 250 km course. Google has hired a lot of the people behind that. These cars "see" in 360º - so they don't see or drive like a human.
Some of these self-driving features are here already - there are cars that self-park, that control your distance from other cars and that warn you when you're going to hit people - and more are coming into showrooms. There are many efforts to develop fully autonomous cars going on - Induct in France is selling cars for campus use. You will not lose the ability to move around as you get older, because of this tech.
Programmers can change the world through this sort of tech.
Mobility on Demand
Imagine pulling out your phone and summoning a vehicle that suits your journey: a mobile office, or a pleasant place to spend time with the family. You need a car that can drive itself around, deliver itself, refuel itself and park itself.
This has big consequences for our city. People ask the car dealer for the car they need for their life. And they choose for the edge case - the biggest demand they're likely to make of it. What if that question becomes "what car do I need today?". Most trips are done alone. That's a very different sort of car - smaller, lighter. It's not just greener than today's car, but it's greener than public transit. And it makes the electric car practical. Robots don't care about how convenient or not it is to refuel. The US would no longer have to import oil. It might break the pesky US habit of going to war in oil-producing countries…
Some states are starting to pass laws allowing this tech in the US. Many countries in the world want this to happen - so their companies can replace the incumbent car manufacturers.
There are downsides:
- Will we walk at all?
- Will people have autonomous cargo carriers following them?
- Will our cities start sprawling more?
- Will we tolerate longer trips?
- The security issues are real.
Minority Report is actually about self-driving cars. Do you want your car to take you to the police - or to prison? And people do not like being killed by robots. They prefer being killed by drunks. There will be accidents. People will be hurt.
This is the chance for the computer to become the most important thing in the car - and that puts the car on Moore's Law curve. Cars are changing really slowly right now. Early adopters are basically stupid people with lots of money. But they can support innovation. This is the opportunity for a bottom-up approach to implementation, rather than via central control. And it'll be surprisingly fast.
Parking & Traffic & Delivery Robots
We are not going to need the parking lots we have now. Replace parking with parkland? That would be nice, but probably won't happen. There are all sorts of alternative uses for this land that beats parking, though. When we have small delivery robots - it becomes possible to get pretty much anything in around 30 minutes. How does that change manufacturing? Retailing? Ownership?
It's time for an Apollo-like resolve to make this happen. This is a hard problem that's become tractable because of advances in technology. You'll see them on the road in the next 10 years - and the change will be dramatic.
Will people still own self-driving cars?
Yes, probably. Rich people, certainly.
*How about the joy of driving? *
Most people don't enjoy their commute. But driving in the mountains? Perhaps. The steering wheel isn't going away for a long time. But will you trust people to drive who don't drive very often? Look at how quickly people drop their car habit of they move to Manhattan.
Will delivery be disrupted fast?
Yes - but the problem is we're developing and testing this with small cars. Also, most of the projects are about saving lives and improving life, not replacing jobs.
Could you talk a little about Singularity University?
It's not really about the singularity, and it's not really a university - but it's great place. But a lot of people have written about the idea of artificial intelligence improving itself. And that's not really what we do either. We're interested in the lead up, the state where technology change is happening so fast that people have problems keeping up.