Post-CES musings on drones

Thanks to Colin Guinn (Hangar), Tian Yu (Yuneec), David Merrill (Lemnos Labs), Lorenz Meier (ETH/PX4) and others for hanging out with me at CES and sparking the thoughts that led to this short post.

DJI Mavic, all folded up!

I made my annual pilgrimage to CES last week. And for the first time in a long time, I went as a civilian observer with no ties to any specific company — which proved to be awesome. With no back-to-back-to-back meeting schedule, I was able to take the show in at an appropriate distance.

Two observations.

I. There are fewer drone companies this year than last — 2016 included a major Extinction Event. In early 2016, the number of UAV companies at CES, especially from China, was enormous.

2015 marked the Drone Cambrian Explosion, with a vast array of drones from a vast array of companies, each drone with curious though mostly minor variations in features, emerging from the primordial ooze. It was a sight to behold.

But just like that, 2016 included the Drone Extinction Event, as many companies seemed to have quit the drone hardware game (or actually folded) as tough, powerful leaders (DJI) crowded out the market with superior products at shockingly low prices.

At this point, Yuneec and DJI seem to be the only credible players left in the high volume drone hardware space. GoPro looks to re-enter this year too, and I hope they understand they must be in it for the long haul, or they should quit now.

While there are a number of companies still set to launch a drone, I consider the door shut to all but the best capitalized, most iron-willed firms. All others are racing into a brick wall.

Put another way, making a drone the user will actually buy today is orders of magnitude harder and more expensive than it was just two or so years ago. The bar has been set, and the bar is high.

II. There were no major drone tech/product announcements — the technological rate of change could be slowing. Interestingly, I didn’t see any major product announcements this year. The 2017 UAVs at CES were mostly minor updates from last year’s.

I believe this is because the must-have features for a UAV are now (mostly) in place. This is a testament to the amazing speed of the market: the sophistication of tech in widely available drones today is jaw-dropping. Today you can purchase a drone with a miles-long HD wireless link, an epically good camera with rock-solid stabilization, and sophisticated sense-and-avoid behavior for about $1000. Crazy.

The rate of change is slowing for a few reasons.

Until recently, drones were obviously missing features that anyone would obviously want. It didn’t take a genius to figure out what the user wanted — but it did take an array of geniuses with a lot of money to actually deliver. These key features were, in order of appearance:

  1. Stable video — shaky video is hard to watch. The three axis gimbal solved this.
  2. A good wireless link — being able to see what the drone is seeing in real-time is a must, and as a user I don’t want to live in fear that I will lose control of the drone if the link flakes out. DJI’s Lightbridge and other standards solved this.
  3. Sense and avoid — I don’t want to crash my drone. Give me piece of mind. Intel RealSense (Yuneec) and DJI’s Movidius/Intel-based solutions are solving this.

Each one of these features took more time and money than the last to bring to market. They are now standard and expected. This means smaller companies won’t be able to enter the game.

It follows that newer features will take longer to develop, so it’s not surprising that there weren’t big new announcements this year. It also follows that the newest features tend to be less desirable than the ones that came before. (I posit that there exists a general law of product maturation that explains this phenomenon: perhaps I’ll cover in a future post.)

By way of analogy: smartphones, when they first came out, demonstrated big jumps in features from generation to generation. Each new feature “wowed” the users — first was obviously multi-touch itself; then GPS and reasonable data rates (3G); then voice control (Siri) and so on. Eventually, the pace slowed down such that today, all phones feel pretty much the same and the new features (better camera, water-resistance) seem less amazing. Each new feature is a little less “wow” than the one that came before.

I argue this same pattern is happening to drones. It’s not a bad thing, by the way — it’s just the natural course of tech products.

That all said, I do believe there are some big new things ahead for drones, despite the trends CES would indicate. We are still early in the drone story.

Two predictions.

I. The stage is set for true autonomy — self-driving drones are coming. For the commercial user, one big missing feature actually involves getting rid of something found in today’s UAVs — pilots.

Just as self-driving cars are eliminating drivers, drone companies will begin to eliminate pilots. The economic pressure to remove humans from the equation is too great (in the commercial domain), and the technology is there — coming from both the drone industry itself, and from the self-driving car industry as well.

DJI’s Inspire 2 has just about all you’d need to eliminate the pilot — sense-and-avoid in all directions, extensibility for adding an LTE link, you name it. We’re close.

As I see it, the only missing pieces for self-driving drones are:

  • Regulatory. BVLOS, or “beyond visual line of sight” operations, requires an FAA waiver. But regulations have a way of playing catch up to the tech and the user’s needs as opposed to blocking innovation completely. Just as Part 107 made commercial drone operations much easier, a new set of rules will allow truly autonomous drones to enter the airspace. It’s just a matter of time.
  • Wireless. Truly pilot-less drones want an LTE link to the cloud so that they can be told what to do from a distance and so that they can return their results to the user. But this isn’t an intrinsically hard technology problem to solve; it will just require cooperation with the carriers.
  • Product/market fit. The hunt for truly nailing the UX and delivering value with a pilot-less drones is ongoing and should not be underestimated. Getting the details right will take time. But it will happen.
Google Wing UAV

II. The VTOLs are coming. To date, the mass market has been dominated by multi-rotors. Amazon and Google (X, specifically) have shown interesting progress with VTOLs (“vertical take off and landing” drones — think drones that can move around and hold position like a helicopter when they need to, but can fly like planes when they want to: this means a combo of the maneuverability/usability of a multi-rotor drone with the flight times and range of a plane). Parrot has made a fixed-wing plane (The Disco) widely available too.

With advancements in open-source autopilot software that can power VTOLs and the increasing sophistication and ambition from major drone manufacturers, I predict that VTOL’s will begin to hit the shelves of (semi-)mass market outlets soon — they won’t be limited to the likes of the military, Amazon and X. They will not be as high-volume as traditional multi-rotors, but a truly usable VTOL will begin to demand the user’s attention and dollars.

There was no explicit evidence for this VTOL trend at CES, but I see it coming nonetheless. Small multi-rotors are not the end-game for UAVs.

We’re getting far enough into the drone saga to see patterns — but big developments and opportunities are out there on the horizon.