Until recently, that wasn't possible. Now, given Moore's law (particularly the exponential curve that shows improvements in calculations per unit energy), it is possible to make a drone fly and swarm like an insect at an affordable cost.
Here's a reference point: In my early days studying Astronautical Engineering, I focused on the design of control systems. We built controls that could turn unstable, high performance jet aircraft and spacecraft designs into stable, usable platforms. To do that, we used expensive computing platforms to make the calculations needed to correct the flight of an unstable jet/spacecraft prototype up to 32 times a second (if we didn't, the system would instantly tumble/fail/become a smoking hole). In contrast, these tiny drones make corrections at a blazing 600 times a second on inexpensive chips.
On top of all of that, the drones currently in the lab aren't merely keeping themselves stable for human control/piloting like most of the drones now in use. They can now control themselves individually and as a swarm.
So, basically, the only barrier to advances in autonomous drone tech are improvements in controls design, made real in software. That means the pace and breadth of the advances in autonomous drone tech will be fast and furious for some time to come.
Today's video is by the head of the UPenn drone lab (I think he should brand all of the videos coming out of the lab as "controls engineers gone wild"). New items to watch for in this video. Drones that can:
- Lift heavy objects as a swarm (payload scales).
- Build complex structures as a swarm.
- Sweep buildings.
- Operate without GPS to navigate and map structures.