I'm really not that sure that drones like the one in the video are more complicated than self driving cars. Self driving cars have to handle a lot of external stimulus, like other cars, road regulations and thrust/steering balances on the road. This helicopter is quite free to move and can be "easily" tracked by gps or such. Only the remote landing can be tricky depending on the situation.
I agree that a self-driving car is much harder on the basis that it has to handle other cars alone.
With that being said, a small drone like this is harder than it looks for a couple reasons.
First, there's not enough payload capacity for a good, accurate IMU. You're stuck using lightweight but awfully noisy MEMS gyros and accelerometers along with drifty barometers.
Plus, there's not enough payload capacity for an off the shelf LIDAR or other long-range depth sensing. That means detecting obstacles needs to be delegated to 2D imagers plus CV algorithms, with some short-range depth sensing. IR ToF like a Kinect 2 or even a grid-based approach like a Kinect 1 might work in a favorable environment, but I wouldn't trust it outdoors even at close range.
If I had to build a pizza-delivery drone, I'd use a bottom-facing camera like the AR.Drone does to try to provide a position reference independent of the bad MEMS gear, and I'd use CV algorithms combined with a last-ditch short-range obstacle sensor (like IR or ultrasound) to attempt to avoid obstacles. Once I got to the destination, I'd delegate landing to a pilot, since "finding the front door" is a surprisingly hard problem.
Multicopters like the one in the video have no trouble reaching much higher altitudes, where obstacles would likely be fewer and easier to detect. Given a specific delivery area, a reasonable cruising altitude could be preconfigured and the flight could probably be navigated by GPS with no CV whatsoever. Then, like you say, a human pilot could handle the landing and subsequent take-off.
The location of the pizzeria could even be chosen such that take-offs and landings at that end of the trip could be entirely automated, by GPS or perhaps with some much more basic CV.
> Self driving cars have to handle a lot of external stimulus, like other cars,
You say that, but you still have obstacles to avoid (trees, tall buildings and hills), "driving" conditions to contend with (high winds) and other "vehicles" (birds, other drones of such a scheme takes off).
> ...road regulations and thrust/steering balances on the road.
Thrust and steering is a great deal harder when you don't have a solid contact (ie movement in air vs wheels on a road). If these drones were to ever be a commercial tool, then they would be bound by regulations as well.
Plus with helicopters you have additional complications that aren't an issue with self driving cars: three dimensions to worry about rather than just two (as you don't need to worry about altitude if your a car), regional air space limitations (eg are you near an air port?) and the obvious issue of somehow notifying the customer that the drone is outside their front door (assuming the AI is smart enough to determine what bit of the house is the front door).
And to make matters worse, each drone would only be capable of one drop off at a time as you have space constraints and trust issues (you couldn't trust customers not to steal other peoples food so you'd need some kind of method to secure other customers food). Where as this wouldn't be a problem with a human delivery boy nor even an automated delivery car.
But that all said, I wouldn't say that these things are more complicated either. I think they're both equally complicated but for different reasons.
> This helicopter is quite free to move and can be "easily" tracked by gps or such.
Cars can also be tracked by GPS. That's how self-driving cars work and what's used in SatNavs. Autonomous tracking is the easy bit as that's been available in consumer technology for the best part of a decade already.
Trees and buildings are static environment and can be mapped. I don't see how bird traffic can compete with road traffic, this comparison don't hold much to me.
The third dimension is only an issue at takeoff and at landing which are by far the greatest show stoppers here. With maybe economic feasibility.
Cars also have road navigation to handle with ugly stuff such as dynamic roadblocks and obsolete maps, all of which don't really apply to flying.
Maybe in a mid/long-term future we'll see regulations for drones but it is very close to free for all as it stands right now
Unless there is some great AI that can handle the landings I think that part can be left to a human. Each situation can be very unique with obstacles such as trees, power lines, other buildings in the way and a human can make better decisions about those on the fly.
Plus if you are delivering to a business in a crowded downtown area you'll probably need to call the recipient while hovering then identifying and directing them to a safe drop off spot :)
As opposed to other drones (we are assuming a future where this isn't a novelty, so lots of air traffic) and regular aircraft, flight regulations, weather and birds.. and yes, the landing and obstacles at low altitude.
Now maybe if they never had to go low but literally drop ship the packages..
