I wasn't very keen on writing about autonomous driving to tell you the truth. I prefer talking about innovative technologies that haven't been widely discussed in popular media. No matter who you ask, they'll let you know that we'll have autonomous cars in the next few years. In this post, we'll look at the different stages of autonomous driving, how close we are to a self-driving world, and what is slowing down the development of fully automated vehicles.
Before we go into technical explanations, let me first say something from a law point of view: The Penal Code and also the Code of Obligations (I refer here in particular to causal liability) have human behavior as their starting point. It will be a great challenge for the courts to pronounce justice in the event of accidents involving autonomous vehicles (also, keep in mind the possible moral dilemmas an autonomous car will have to solve).
Alright, I will stop and give you a break (and show you how I picture myself being chauffeured around while writing for this blog).
The underlying technology which enabled any thoughts of really moving forward with this technology is machine learning.
Specialists have characterized five levels in the advancement of independent driving. Each level depicts the degree to which a vehicle assumes control over obligations from its driver, and how the vehicle and driver interface. I'd like to go briefly through the five levels of vehicle automation:
Driver Assistance: these systems are designed to support the driver. However, they are not capable of taking over control from the driver.
Partly Automated Driving: systems can take over control, the driver remains in charge of operations. This is the case for applications that facilitate parking.
Highly Automated Driving: the driver can disengage in certain situations and let the car drive itself for extended periods of time.
Fully Automated Driving: the car drives most of the time independently, while the driver can lean back and read the newspaper. There might be a situation where the driver needs to be capable of driving.
Full Automation: the vehicle assumes all driving functions, the people in the car are passengers.
While there are many systems for assisted driving (such as lane-keeping, etc.), we are still a long way from fully automated cars. Why is that?
The automotive industry has overrated the pace they will be able to provide consumers with autonomous vehicles, attributing this to human behavior. What they mean by this is, that streets can detect objects, lanes, and obstacles of any kind, however, the tricky part is figuring out what actions the car must take after identifying such an object. Autonomous vehicles can sensor, film, and, therefore, see what is going on - but how can they make sense of it? With current technology, this seems to be the biggest obstacle.
Similarly, testing is showing self-driving vehicles the better purposes of driving, now and again known as “micro maneuvers.” If a vehicle ahead is moving slowly searching for a parking spot, it is ideal not to follow too intently, so the car has space to once more into an open spot. Or then again to realize that if a vehicle is pushing out into an intersection, it very well may be a sign the driver may dash out regardless of whether he doesn't have the option to proceed.
The technology is now accessible to make a vehicle that won't hit anything. In any case, such a vehicle would continually slam on the brakes (what a nuisance!). In order to progress more quickly with autonomous driving, a solution could be to have autonomous vehicles on the roads in controlled settings and situations (e.g., shuttles for a given route).
However, even if there are a lot of systems in place, we are still quite some time away from a self-driving world. How do you feel about that? Let me know in the comments or let's discuss on Twitter! As always, below are my sources and, stay curious!
Furthermore, I have used this New York Times article. Enjoy!
You can find more on the different levels of vehicle automation here.
For moral dilemmas, a self-driving car could face and let engineers and scientists know how you'd decide: MIT's Moral Machine.