You can pull certain tricks with ballistic delivery methods. A depressed trajectory (flatter profile, lower apogee) would give you the option of coming in through a window, with an additional benefit in that it would be harder to intercept the payload.

You would also need to limit this strategy to payloads that can withstand shock loads of multiple hundreds of G (soups and hard cheese, maybe even bread adequately packaged. may need a different method for floral deliveries).

Proof of concept shouldn't be too difficult, most of the tech already exists (Raytheon and Northop-Grumman have had a handle on it for decades). DHS might not be too amused, however.

> would give you the option of coming in through a window

You're thinking too small. I want my uber eats delivery to land directly on the plate in my cupboard before I even set it on the table. Please provide appropriate ballistic modelling for this thankyou

I'd pay for additional convenience of food flying directly into my mouth.

Move fast, break teeth.

Miley Cirus could do promotion work for such delivery methods.

Pure comedy. Thanks.

The catapult launches the drone, not the package. The package is delivered as normal and the drone is then returned and "caught" with a pretty impressive hook system for redeployment.

The package is not delivered "as normal" - it's dropped via parachute. The parent poster's point is that delivering things via parachute comes with all kinds of risks in comparison to targeted landings (ex. wind gusts during drop, power lines, blowing onto a roof, parachute failure causing breakage, etc.)

These are similar risks to the Google Wing concept. The parachute has fewer risks of collision with the vehicle since the vehicle is nearby for a much shorter period of time and is not making a flightmode transition.

The issue is simply a matter of minimum required landing area. Hovering could be a little smaller than parachute delivery (16 ft), but still must have a "safe area" is approximately similar because the vehicle must hover without nearby obstructions.

From what I've seen, Zipline is being run between managed locations, rather than delivering to random points on the map, but I could be wrong about that, I last read about it a couple of years ago.

Also, like a lot of solutions, this doesn't have to be perfect, it just needs to be better than the alternatives, which Zipline appears to be in the places it is being used.

Wouldn't variations in wind be a non-starter for catapults?

The catapult launches the drones into the air, which then parachute the package to the location.

Real Engineering [1] and Wendover Productions [2] have really nice videos about their system currently deployed in Rwanda.

[1]: https://www.youtube.com/watch?v=jEbRVNxL44c [2]: https://www.youtube.com/watch?v=bnoUBfLxZz0

That's just incredible. Thanks so much for sharing.

As long as you launch an object capable of aerodynamic control (glider), you can compensate for wind by launching with excess energy/speed. Launching could be easy by a catapult of a couple hundred meters (50m distance for 100m/s speed gives 10g, 200m distance for 200m/s gives those same 10g.

My best idea would be a flywheel, because for 5kg mass (including a sled/rope), and assuming force-limited operation, you'd need 100kW peak power at the end for the 200m long 200m/s exit speed installation.

Spin the flywheel, and engage a clutch to connect it with a winch that pulls the sled via a (Dyneema) rope. There are magnetic clutches that provide precise control over the force, allowing it to run the glider/rope near their structural limits.

Rough, back-of-the-envelope calculations suggest viability of 25m long catapults that launch suitable freight at ~200m/s, but "suitable" here means ~100g, likely requiring fancy rigid foam fillers for the box and limiting to sufficiently solid freight.

If you figure out how to dig a 100~400m deep, 2~3m diameter hole cheaply, you could save massive horizontal space. Launching out of an elevator shaft of a highrise or on the wall of a tall building could be practical, too. Just drag a thin rope up behind the sled, and use it to pull the main rope back up. The sled should be far too light on it's own. Click/hook in the next glider, while getting the flywheel back up to speed. This should allow 10~20s spacing on a single sled/guide, and up to ~2s if you robot-mount the glider fast enough to the next sled on an array of parallel guide rails. As soon as one glider clears the launch space, the next one could be moved in. That'd be important to the economics of building / converting a tower for such a use case.

And even if the contents require long (and thus horizontal) catapults, it might be feasible to launch the empty glider from much smaller (50m vertical would also solve most clearance issues) facilities to use most of the range for actual delivery.

The catapult launches the drone, not the package.