How a sphere flies.

Conventional drones hang their rotors in the open air because that is the simplest way to move a lot of mass. Orb One takes the harder path: the lift fans are meant to live entirely inside the TrueShell, with the only outward sign being four pairs of hairline seams that arc across the surface. Each seam is a disguised channel – one side draws air in, the other routes it out – so what looks like a part-line on a premium earbud is actually the airway. We call this architecture SeamRoute.

The physics here is real but unforgiving. Ducted and shrouded fans are a proven idea, and enclosing the airflow can cut blade-strike risk and tidy the acoustics, but ducting also adds internal drag and weight, and narrow intakes limit how much air you can move. That trade is the central engineering problem of the whole product, and it is the main reason Orb One is framed as a concept targeting 2030 rather than something we claim ships today – the efficiency has to come up before the flight times become honest.

What SeamRoute is designed to buy is not raw performance but a different feel: a flying object with no exposed propeller, no cage, and no obvious front, that you could catch in your hand without flinching. We would rather under-promise the endurance and get that experience right than bolt visible rotors onto a sphere and call it seamless.

Orb One is built around the idea that the safest, nicest thing a small drone can do is take care of its own takeoff, hover, and landing. SoftSet is the autonomy behavior we are designing to handle those moments: it lifts off on its own, holds a stable position, and when it is time to come down it decelerates, looks for a clear spot, and sets the sphere down gently – or lets it roll to a stop in your hand. The goal is to remove the two phases where casual pilots crash most: the launch and the landing.

Autonomous position hold, return-to-home, and obstacle-aware descent are established techniques on today's better drones, so the building blocks are real. What we are targeting for 2030 is doing them inside a fully enclosed sphere that has no GPS-friendly open frame and only one sensor band to work from, reliably enough that catching it in your hand becomes a normal action rather than a stunt. That reliability bar is high, and we would rather narrow where SoftSet is trusted than over-state it.

Manual piloting is meant to sit on top of this, not underneath it. You could take the stick to move the sphere where you want, but the autonomy stays in the loop – holding steady when you let go, refusing to fly into something it can see, and always able to take the landing back. Remote piloting is for intent; the machine is designed to keep the part that has to be safe.