Not_airplanes with not_pilots...

[elaw]

...sure are not_slow! [urlhttps://www.youtube.com/watch?v=fBiataDpGIo][/url]

[Not_Karl]

Can't wait for them to rebel against us...

[ocelot]

Note though that it has the course trained into it, so it wouldn't be able to cope if things moved and humans probably can.

That said, I do want to see more of the video from the onboard cam...

[3WE]

There is software that makes an autonomous Quadra copter hover stability, or deliver Amazon packages.

Optimizing a a race course with turns, climbs, descents…a much smaller challenge.

[Not_Karl]

Optimizing a a race course with turns, climbs, descents…a much smaller challenge.

Identify and terminate all humans... even simpler.

[elaw]

Identify and terminate all humans... even simpler.

Are you a Dalek?

[elaw]

There is software that makes an autonomous Quadra copter hover stability, or deliver Amazon packages.

Optimizing a a race course with turns, climbs, descents…a much smaller challenge.

Do you really think so?

The amazon/hobby drone just needs to climb to an altitude high enough AGL to clear any obstacles, travel in a straight line to its destination, drop down a bit, possibly deposit something, then reverse the process, all at its leisure.

The racing drone needs to fly & maneuver at maximum possible speed*, often at attitudes that would be considered "departure from controlled flight" in a human-piloted aircraft, and travel through several tightly-defined spaces between obstacles, all on a very 3-dimensional course. I'd think that would be harder.

* And not necessarily in the sense of "open the throttle and go"... I think in some cases intelligent reduction of "throttle" might be helpful in order to make tighter turns and thus traverse the course faster?

[elaw]

That said, I do want to see more of the video from the onboard cam...

Just make sure you have a barf bag handy when you watch it!

[ocelot]

I've never been particularly susceptible to that

[3WE]

There is software that makes an autonomous Quadra copter hover stability, or deliver Amazon packages.

Optimizing a a race course with turns, climbs, descents…a much smaller challenge.

Do you really think so?

The amazon/hobby drone just needs to climb to an altitude high enough AGL to clear any obstacles, travel in a straight line to its destination, drop down a bit, possibly deposit something, then reverse the process, all at its leisure.

Navigation: No.

But; hovering stably, dealing with turbulence, does the wiring and seprate motors and propellers have any anomolies, and the flight sensing instruments...I consider it pretty damn awesome that all of that can be integrated into a drone that if fly with my iPhone.

Conversely, I expect that a geek, undergraduate damn-average aeroengineer (or race car engineer) might be able to suggest math and code to optimize turns.

Now, don't get me wrong- I doubt it's "turn-key" easy, but what I am saying is getting a quadra copter to fly, flip, rotate, turn, accelerate, stop...all of that seems a bit more impressive than running speed and radius and centripital acceleration (and climb and descent) to beat a human...

...Not_as tricky.

PS: Does the Amazon drone not_need to see and avoid trees and porch roofs and people and dogs and cats and neighbors with shot guns?

[ocelot]

The thing it's doing that's new/interesting is tracking where it is (relative to pieces of the course) optically, in real time. That is not_so easy.

[3WE]

1. new/interesting

1. Indeed.

2. Not_saying it’s EASY

3. The AI/learning aspect, may, indeed, be complex.

BUT JUST SAYING:

An algorithm to iteratively optimize pitch, bank and power is less complex than integrating 4 motors, 4 torque points, 4 lift points, radar altimeter, attitude, and pitch, roll and yaw axes, the effect of airspeed on leading and trailing helicopter blades, game-level-do-this, input from an iPhone attitude indicator AND phugoid management…

This optimization is a logical, next step, that engineers will crack quickly.