Flies are far from simple. Read up on their biomechanics sometime. They have six degrees of freedom--forward, backward, up/down, side to side, and rotating around all three axes. They're fantastically sophisticated flying machines.
Depends on how sophisticated you want the avoidance to be!
On one end of the spectrum, you could have the Roomba-style drone that physically feels the obstacles. Contact on a bumper would instruct the drone to stop, avoiding a flight-ending collision. It would probably need to fly pretty slowly though.
On the other end of the spectrum, you could match the incredible FPV drone pilot Charpu [1] with onboard AI. Flying full speed through very narrow gaps with a long term goal in mind.
Flies fly in far less cluttered environments than drones do (relative to their size and manoeuvrability). Something like a wood pigeon would be a better example. Still, it's not something that should require an AGI. :P
What we really miss is an ADS-B low-cost variant for drones. Safety requirements are different if ones takes into account the unmanned characteristic of drones. Furthermore it will be a great exercise in solving the known problems with the existing version. The main idea of broadcasting navigation data is worth the shot to optimise even more the use of airspace. Without lives at risk technical iteration can be, at least in theory, faster. We need more IETF-style drafts and RFCs rather than classic standards (i.e that require often more than 10 years to reach the market) to see a clever use of such a technology in our daily-routine.
That's available. ADS-B receive only in a 6 gram package. A full ADS-B transponder in a 20 gram package.[1] Now your drone is visible to ATC and aircraft, and in turn, can see other aircraft, in areas where there's ground radar.
This is more for commercial operations in class B airspace, where you need to get permission from the controlling FAA tower. Real estate photography in a city or shooting a movie, for example.
This doesn't help with avoiding stationary objects; that's a separate problem and requires sensors. ADS-B just tells all the players in controlled airspace where the other players are.
Yes I agree with you but you are point out the existance of low weight packages while I would say that cost is a more important constraint for drones. I guess users can argue with the idea of equipping their low-cost drone with an expensive certified ADS-B receiver. It won't be as cost effective as one may think.
Also one has to consider that the mode S transponder is using a 24bit identifier. While this has a lot of sense for modern civil aviation (there are not as many planes) the same cannot be guaranteed as soon as you have many more temporary users of your airspace (e.g. swarms of Amazon delivery drones). IMHO regulators should aim first to create an effective solution that works well for drones (e.g. allowing ATC to broadcast digital commands to guide drones if necessary, btw there is a similar effort already for commercial planes) and then find a smart way to integrate commercial and drone aviation systems.
This isn't just a problem for drones. It is an ongoing problem for gliding (hang and not), ultralight aviation, not as light private aviation and so on.
The aviation powers that be are simply not interested in a standards based, low cost, uncertified location beaconing system with low power consumption even through the world desperately needs such a thing.
ADS-B is not drone specific, but it is very useful to make drones more safe. The problem in the real world is that many pilots of small aircraft refuse to use ADS-B. For some, there is concern about the government tracking their movements, and of course there's extra cost for extra equipment.
There's also concern already that it's way too easy to spoof ADS-B with false data, and that it won't scale well to the number of drones expected to be flying.
And we have self driving cars, arguably far more complex than the relatively simple task of flying through empty space.