Skydio: $1k Skydio 2 drone launch takes aim at DJI

To be fair, there are strict FAA rules that tons of people completely disregard. It seems like it'd be in the companies interest to do its best to ensure their market doesn't collapse due to further regulation.

Seems like it could be an attempt to understand and police some of that activity rather than "appalling snooping".

We are working with various commercial and industrial users, but don't currently have anything to announce regarding a consumer accessible API (partially so people don't do stuff like make weapons). If you think you have a really good application, please get in touch.

https://medium.com/skydio/skydio-2-drone-industry-34b5634d48...

This is one of the reasons we don't let the drone keep following for very long if the connection is lost to the phone or beacon controlling it. As a person operating the drone, it's your job to know local and national regulations and follow them.

Thin objects are really hard, and are probably the single area we have made the most software progress since the R1, especially very thin twigs and branches (it's still not perfect, fly carefully). We completely rebuilt the stereo vision algorithms to try to handle those better, basically by incorporating semantic knowledge (recognizing what is a tree, or a power line) in addition to traditional stereo vision techniques. To my knowledge, our systems are significantly more advanced than even the state of the art in academic research papers, so I would be impressed if there are any other drones that match our performance today.

We have a 3-axis gimbal, the footage comes from the front camera.

Pics: https://www.theverge.com/2019/10/1/20892377/skydio-2-drone-a...

Take a look at the warranty. If the drone crashes, it's our fault not yours. In that sense, flying Skydio 2 should require about the same minimum skill as using a smartphone camera (i.e. none). We say a three year old can operate it, but that's not a hard minimum ;)

That said, there is always a skill ladder to climb if you want jaw-dropping professional footage, where understanding lighting, composition, editing, etc. benefit the quality of what you can make. The drone is pretty good at autonomously picking good shots, but there's also a lot of manual control possible, especially with the controller.

Depends on what kind of drone. Flying a FPV / racing drone requires some skill (but not much if you've played games). Flying acrobatic drones in third person (eg. https://www.youtube.com/watch?v=duHCDwH9G2M) requires a lot of skill.

Compared to those, flying drones like those from DJI require almost no skill. These drones do some form of obstacle avoidance on their own. You basically have to tell them where to fly and they will go there. Some are smarter than others in how they execute the manoeuvres, but some common sense will get you a long way with those.

It looks like the point of this product is that it flys itself.

All of our development (software, mechanical, electrical, marketing, etc.) is done at our headquarters in Redwood City, CA.

The parts come from all over the world, I don't know the exact breakdown of country of origin. Assembly, testing, and shipping all happens from Redwood City. Eventually, it might become necessary to move some of that work overseas, but we really enjoy the rapid development and iteration speed that comes from having the assembly line across the street from the engineers.

Yes, it works just fine over water. If you are going to use it for surfing, I would recommend taking a beacon in a waterproof bag of some sort, just to have the extra range and make sure it doesn't lose you. Keep in mind that the warranty does not apply to water damage, so fly at your own risk.

Also, make sure you can get back to land when the battery runs out, or that you're really comfortable hand-catching it.

Edit: I have been informed that you can manually set a "home point" on land to return to when the battery gets low. Also, apparently the surfing footage we captured was done with a pilot on shore with the joysticks as a range extender, and the subject being tracked purely visually (no beacon involved)

It's really hard to see "through" propellers with a camera, and we need to mount our top cameras on the front arms. That's one of the examples of why we think that great autonomy is possible only if it's designed in from day 1.

Thanks!

It is always good for DJI to have some competition. They have been spending too much time going after GoPro cameras instead of advancing their quadcopters.

DJI makes multiple drones at multiple price points. I think not making it fold-able is going to hurt them. I know some people use their drone to follow them, but a lot of people just to fly their drone. I don't see why they would have any advantage in that space. I feel this probably cool technology without a good market fit.

Yes, we pick the cool looking stuff for publishing, but the bigger factor is that we really do program the AI to fly in cinematic ways. Because it has a full 3D model of the world, we can actually model in software a lot of what cinematography traditionally does manually.

gotta start somewhere!

Somewhat related, any plans for integration with VR headsets?

This could either be first-person flying, or the generation of stereo-correct video (using offset cameras) for later viewing.

In other words, you're unlikely to bump into something you're moving away from, which is why humans don't have eyes on the back of our heads either.

This has eyes mounted everywhere, but I presume it dedicates its visual cortex (to stretch the analogy a bit) to the ones facing in the directions that matter.

Thanks a lot for your reply!, yeah I was thinking 4k * 6 cameras is an awful lot of data to process.

You may want to add a "Show HN:" to the title