Skydio R1 Autonomous Sports Drone In-Depth Review

Skydio-R1-Drone-Main

My quest to find a drone that can follow me while cycling and running may finally have found its maker. After years of trying all sorts of products, the Skydio R1 came along. And it did so in a big way – packed with 13 cameras and enough obstacle avoidance tech to autonomously navigate even the craziest of running trails, I was eager to put it through its paces. And there’s been no better place for that than the coastline and outback-like trails of Western Australia.

Thus, for the last two and a half weeks I’ve been letting this little buzzing bumble bee follow me around while running, road riding, mountain biking, and an assortment of other random activities.  Is it finally the drone I’ve been looking for?  And if not – what’s it lacking?  I’ll dive into the good, the bad, and the ugly – as I always do.

Now, if you want to get straight into the action on seeing just how well it does, here’s my full in-depth review video. This also includes all piles of footage from running, road cycling, and mountain biking.

Note that the company sent me a media loaner R1 drone to test out. But that’ll go back to them in the next couple days (since I seriously don’t have any more room in my luggage for the trip back to Europe next week).  That’s just the way I roll.

With that…let’s begin.

Unboxing:

Now I’ve put together an entire R1 unboxing video – which also includes some other little snippets as well like size and weight comparisons.

x

I even took an entire set of unboxing photos, only to find out the SD cards was apparently not inserted into the camera.  Thus, the images were instead transmitted into thin air.  No worries, the lighting for the video above was better anyway.

The Basics:

Skydio-R1-Camera-Gimbal

Now that we’ve got everything all set, let’s dig into some of the basics.  Obviously the main goal of this unit is to follow you while you’re (hopefully) doing something cool, ideally sports-related.  I’m sure there are use cases for non-sports specific things, but let’s be honest: This is all about sports.

As I’ll show, it’s not terribly awesome at being a more general-purpose drone.  Part of that is because the R1 is only controllable via your phone (Android or iOS), which uses WiFi to control it and thus is range limited.  You’ll want to download the free app and get the drone all registered.  It only takes a few moments to complete this process.

Skydio-R1-On-Phone

If you haven’t done so yet, you’ll also need to charge the batteries.  As noted above the unit includes a USB-C charging cable to a small USB-C dock for the battery.  I’ve found I can usually charge an R1 battery from close to 0% up to 100% in about 35-40 minutes.

DSC_2114

What I like about the use of USB-C here for everything but the small charging plate is that I can use that USB-C cable for other purposes.  For example it charges both my 2016 MacBook Pro, as well as my Lenovo T470S laptop (both via USB-C).  Inversely, I can use my laptop chargers to charge the drone battery plate.  The only thing that would make this better were if the battery had an actual USB-C port itself.  But yeah, I’ll take the current setup for sure as one less charger I have to take with me on a trip!

DSC_2118

The next piece to understand is the units 13 cameras.  Twelve of these cameras are primarily used for obstacle avoidance, and ring the outside of the drone (two in each corner), as well as have two each on the top and bottom.

DSC_2121 DSC_2119

On the bottom those two cameras are protected by a small set of rubber feet, which act as your landing gear.

DSC_2123

Meanwhile, on the back of the drone is where the battery is inserted.  I’ve found it consistently gets about 13-14 minutes of flight time, which is pretty darn close to Skydio’s claims.  What’s also nice is that I can actually fly the battery down to 1-2%, versus having a forced landing at 10%, or endless alarms below 20% like some other drones.  I can easily tap the batter button to get a light ring showing me battery status.

DSC_2124

The unit has four props, all of which are protected by the frame of the drone.  The frame of course serves two purposes.  First is to protect the props from things it might hit (or to protect those things, like humans), and second – more importantly, is to house all of the obstacle avoidance sensors.  There are two ‘types’ of props on the drone, one type of which can be identified by the silver ring on the inside of the prop.  Whereas the other type of prop doesn’t have such a ring.  The R1 includes a full extra set of props in the package.

DSC_2120

Back on the app, when you power on the drone it’ll power on its WiFi, which in turn you’ll connect to via the app.  Basically, the same as most other WiFi controlled drones.  Again, there is no remote controller here.  Once connected the app doesn’t have a ton of menus to dig through.  Things are basically divided up into three categories: Videos (watch past recordings), Fly Now (to fly it), and Settings (settings, obviously):

2018-03-21 16.34.29 2018-03-21 16.34.20 2018-03-21 16.34.32

Looking at the settings first, there’s honestly not a ton of options in here. You can essentially change the video quality or change how you move the camera manually…and that’s about it.

2018-03-21 16.34.32 2018-03-21 16.34.41 2018-03-21 19.02.10

Next, there’s the Fly Now option, which is where we’ll start a flight.  Once we tap this it’ll connect to the drone and show us the current camera view.  We’ll also have the option to hand-launch or ground-launch.  In general you would ground-launch if you’ve got nice clean ground (like pavement), whereas if you were in snow or sand, you’d want to hand-launch (to avoid crap getting into the cameras/props/etc…).

2018-03-21 16.34.17 2018-03-21 16.35.55 2018-03-21 16.36.10

Hand-launching is merely holding the back of the drone (basically, the battery), and the drone will just gently fly away from you.  Since the props are protected by a casing, the only change of injury here is if you stick your fingers up/down into the props.  Obviously, one can do that if they so desire – but you can also stick your fingers in a blender too.  Your choice (the blender will hurt more).

DSC_2113

Most of the time I just ground-launched, but I did a few hand-launches as well when I was on sand (and there were a few trail cases where I brain farted and wished I had hand-launched).

vlcsnap-2018-03-22-15h36m28s423

Once you’ve taken off you’ll see modes along the bottom that you can swipe through.  Each of those modes is mostly a different tracking option.

2018-03-21 16.36.35

They are as follows, all of which (except as noted) work as you move, so all are follow modes

Orbit: Rotates around you continuously
Side: Stays to your side, like a profile shot
Lead: Stays in front of you, facing you (head-on camera shot)
Follow: Stays behind you, facing you (from behind shot)
Tripod: Stays put, but the camera follows you (like a static camera)
Smooth: Smooth’s out the shot a bit to be less responsive
Stadium: An above-shot, higher up
Joystick: Manual control of the drone

In addition you’ve got the ability to change the drone’s height or ‘zoom’.  The zoom is simply how close to you the drone stays.

The only criticism I have here is that if you’re outside in bright sunny conditions, even on the brightest iPhone X screen settings it’s virtually impossible to see what the non-selected modes are.  By that I mean, if I’m on ‘Follow’ mode and want to go to ‘Lead’ mode, do I swipe left or right? I can’t see those other modes because they aren’t bright enough.

Most of the time I used either ‘Follow’ or ‘Lead’ mode.  Follow mode is typically the easiest because it doesn’t have to ‘guess’ where you’ll go, especially in trails.  That said, I’ve actually had really good luck with lead mode.  But more on that in the Sports Tests section.

The ‘joystick’ mode can be used for general flying around, ala a typical DJI drone.  This works fine, and is much better than the default layout that you’d see in the other modes (seriously, I tried using those controls in the default modes and wanted to throw the phone at the controller).

2018-03-21 17.03.13

Still, you have a practical limitation of about 60-80 meters of distance between you and the drone (the specs say about 100m, but I found the signal cuts out sooner).  Whereas I’ve found with DJI’s at 100m of phone-controlled distance (and many kilometers with a remote control), the signal is generally pretty solid.  This distance limitation isn’t really noticed much when you’re tracking yourself in a sport (because you’re always closer).  But it’s brilliantly obvious as a limitation if/when you try to use the drone for general cinematography, since you’ll quickly run up against these invisible glass walls.

Similarly, the drone has a maximum height of 20 meters.  That’s a bit low in the drone world, though realistically for follow-me type situations, that’s actually pretty high. There’s very little from an interest standpoint when you’re tracking a human and the drone is anywhere near that height.  The wider angle lens on drones makes humans even at just a handful of meters seem far away, being 20 meters high you might as well be in a commercial airplane from a human-focused shot standpoint.

From a footage standpoint your choices are pretty simplistic: 1080P/30FPS, 1080P/60FPS, 4K/30FPS.  This gives you slightly less options for high speed footage like 120FPS found on the DJI Mavic Air, but still covers you at 4K/30FPS. For what I do, I tend to shoot 4K/30FPS for everything, so it fits well.

I’d point out that I didn’t find the image quality all that great on Skydio’s marketing sample videos, but the actual unit I now have seems really quite sharp and clear (as you can see in my sample footage in the video).

Now we’ll get to all the sports tracking pieces in a moment in the next section, but for now, let’s move onto the last of the three tabs – the video piece.  It’s here that you can view lower res videos of your flights.  By default it streams/saves a copy of your flight in low-res to your device, along with the audio recorded on the phone (not the audio from the drone).  You can then take snippets here, including grabbing photos from it.

2018-03-07 15.03.31 2018-03-07 18.59.49

If you take a snippet and the drone is powered on and connected via WiFi, then it’ll download the full high resolution piece.  In addition, you can also livestream your session over 720p at 15FPS. But that’s kinda low resolution and frame rates, so I think it’s unlikely most folks will do so.

Now Skydio seems to have put in a fair bit of effort to having you offload via USB-C connection instead.  For that you’ve got two basic options:

A) Attach cable to your computer, shows up like a USB hard drive
B) Attach USB hard drive directly to drone, offload footage using app

Both methods work really well.  Remember, your limited to the 64GB of internal storage in the R1, so if you’re out for a few days of shooting without a laptop to download the footage, then you might want to use the USB hard drive option instead.  Note that it doesn’t need to be a USB-C drive per se, but you do need to attach it via USB.  For example, if you have a standard SATA external drive, the cables are almost always detachable.  Here you can see that I picked up this USB-C cable off of Amazon for $8, and it makes my standard external drive USB-C.  Or, you can just buy a USB-C drive too.

DSC_2069

You’ll then simply use the app to offload the footage (or footage + logs), using the ‘Sync to USB Storage’ option. At the end of which if successful it’ll delete the contents from the drone.  The 64GB of storage gets you about 1.5hrs of usable space at 4K/30, or 4.5hrs of video at 1080p/30.

2018-03-21 19.01.58 2018-03-21 19.02.10 2018-03-21 19.02.17

In addition, you can use either of the two included USB cables to connect the drone straight to your computer, for which it then appears like any other connected hard drive.

Note that it does require you power up the drone to do this, and I found the drone got pretty darn hot, but that never seemed to impact anything.  Also, it doesn’t seem to use much battery. I downloaded an entire full drone’s worth of footage in about 45 minutes, and only lost about 1 bar of battery (so about 10-15% or so).  I will note the throughput speeds on the drone via USB-C seem a bit slower (even on a few faster drives I tried). So offloading a full drive is in the order of 40 minutes or so, and sometimes it also seems to give a failure message at the end even when it seems to have otherwise succeeded.

Also of note here is that when you export out via cable, it’ll actually grab the audio file from your phone and stick that in the files as well, so that’s kinda cool as they perfectly match up. You can see those below. The .MOV file is the video file, and the .M4A file is the audio file. The MP4 file is a small preview variant of it, and the rest are various Skydio log files.

image

At this point you’ve got all your files like any other camera or drone and you can do as you please with them.  So now let’s head back and talk specifically about sport tracking.

Sport Tests:

vlcsnap-2018-03-22-14h50m14s223

With all that background out of the way, let’s head outside and get it up in the air.  It’s not until you’re in the air that you select the various follow modes.  Once launched though you’ll either want to situate yourself somewhere in front of the camera, or simply rotate the drone to point at you.  After you’ve done that you’ll see a little icon hovering above you:

2018-03-21 16.42.45

This means that it’s now found a viable object to track. Basically, a human. It’s not capable of tracking cars or buses or dumpster fires. I haven’t tried a dog or a cat or a polar bear. Perhaps those would work, perhaps not.

To track that object, you’ll simply tap it, and it immediately goes into follow me. As it’s tracking you, if looking at the screen you’ll see that it makes this little rotating hoola-hoop around you, indicating it’s got a lock on you.  Even when you’re really far away, this is still going on.  You can now wander as you see fit, and it’ll follow you.  Nothing more to do. It’s that simple.

2018-03-21 16.43.10

The tracking is largely done using the singular 4K camera up-front, whereas the other 12 cameras are used for obstacle avoidance.  Those cameras are looking outwards ‘tens of meters’ and trying to determine what you’re going to do next, and how to avoid whatever might be in its way.  In fact, this is what those cameras see:

vlcsnap-2018-03-22-16h06m03s236

What you’re looking at above is rendered using the log files from those cameras.  It was an advanced logging mode that I was able to get the Skydio folks trigger on my unit, which in turn spit out crazy mounts of data (17GB for 6 minutes!).  But from that they were able to generate imagery showing exactly what everything looked like from the drones perspective.

In some ways it’s actually kinda mesmerizing to watch. Here’s a full 6 minute run starting and ending in the same place after tree tunnels and all, in that mode:

Speaking of tree tunnels, I was fairly impressed to see it was able to follow me while running through this tree tunnel.  That’s incredibly impressive – and something no other drone on the market can do.  For example, if you tried doing that with a DJI drone, it’ll just stop at the entrance.  I tried it multiple times – no luck.

Inversely, if I used something like the AirDog ADII or Staaker drones, I could have it actually track me as I ran through the tunnel, but it’d be unable to see me in the tunnel – instead, I’d have to have it well above the trees like a police helicopter hovering near the scene.

But what about cycling?  No problem – I started off with road cycling, trying that out first before mountain biking  Here, it worked exceedingly well in ‘follow’ mode, easily tracking me from behind.

vlcsnap-2018-03-22-16h11m12s492

I then switched to ‘lead’ mode, and it worked well.  In fact, I think some of the coolest footage I’ve got on the unit be it mountain biking or road riding was actually in lead mode, especially when I brought it nice and close.

vlcsnap-2018-03-22-16h10m23s468

In lead mode all was fine until the point where it got off to my side a bit and then got caught up by an incoming shrub.  My suspicion here is that it was trying to get back in place on lead mode as I went faster, but just couldn’t do it (thus, defaulting to side).  Unfortunately, by time the shrubbery occurred, it had no option but to hit the brakes to avoid it – thus losing me.  I show this in the video at the start of the post.

vlcsnap-2018-03-22-16h12m28s638

Still, that was rare.  Keeping in mind that I can lose a DJI drone using Active Track in a matter of seconds on an empty road in a desert, it’s impressive this can track until the battery dies.

Do note that one challenge of the Skydio R1 drone at this point is the maximum tracking speed limit – which is 25MPH.  once you go above that speed the drone will simply stay at 25MPH and track you as long as possible.  But ultimately you’ll escape the view of the drone and it’ll stop and hover there, waiting for it’s next comment before eventually giving up, and finding a ‘suitable place’ to land (you have 5 minutes to get back to it before it lands).  I tried that on a hill section, and it did just that.

vlcsnap-2018-03-22-16h17m17s954

But what about increasing difficulty beyond just speed? Let’s go mountain biking.  First, we can try something more complex than road biking but less complex than dense bush.  In this mountain bike park I roamed around for 15 minutes…twice (two batteries worth).  No issues, it tracked impressively well, despite brief seconds where it lost me behind trees, it’d find me again and all was well.

vlcsnap-2018-03-22-16h18m35s044

So then we’ll head into the real test – the bush-filled mountain bike trails of the Australian Outback (Note: No relation to the Outback Steakhouse).  With two fresh batteries I was ready for an afternoon of tracking.  I found myself a bit of a wider dirt road of sorts to start with, and began my trek:

vlcsnap-2018-03-22-16h19m46s050

Initially, all was well.  So then I took a smaller and twistier single-track trail, and shortly thereafter the drone whacked right into a tree. Here’s what it looked like:

Now, astoundingly, it did not crash.

In fact, Skydio warns that smaller braches (like those on a leaf-less tree in winter), might not be picked up by the optical sensors, and to exercise caution around those.  But a full tree trunk?  It should have caught that.

Still, since the blades are protected, it just sorta bounces off.  And over the course of the afternoon I’d clip a few smaller braches here and there and it just flew right threw them.

But that wasn’t the main issue here.  Rather, it was the R1’s tendency to be too cautious.  See, the R1 wants to have a 1-meter ‘bubble’ around it for passage through areas.  Sometimes it’ll squeeze through in less than that, but that’s it’s goal.  And with all of the various hanging braches and trees on this particular trail, it would often get stuck in areas that I knew full well it could follow through, but it was like a scared first-time mountain biker not wanting to take the steep line.

vlcsnap-2018-03-22-16h21m25s887

It would then just simply wait there until it found another way around (basically you had to go somewhere else), or until you manually took over control and found an alternate route.  And as such, in these specific trail conditions, it made the R1 largely unusable.

Of course, each trail condition will vary.  My guess is that if you were up in the Pacific Northwest with larger pine trees that tend to be a bit more spread out than wannabe-jungle of Australia, it’d probably do just fine.  It doesn’t need a lot of space as evidenced by the tunnel…but it just needs slightly more space than this particular trail gave it.  Plenty of tree skiing videos people have posted show that.

But let’s talk about a few other things related to sport use.  First it should be noted that unlike the DJI solution, you can actually lock your phone and stick it in your pocket and it’ll track you.  Whereas with DJI it’ll immediately stop the drone when you do so.

In my case, for cycling though I found it useful to keep it on my handlebars. I used the Quad Lock mount that I mentioned a week or so ago.  That worked well enough. But to be really clear – you can leave your phone in your jersey pocket locked (or running shorts). I did both. This is notable because if you try doing the same thing with a DJI drone, it immediately stops flying as soon as the screen is locked.

Skydio-R1-On-Phone

What didn’t work well was trying to control the drone using cycling gloves – or even sweaty fingers, on the controller.  Ultimately, what Skydio needs is a simple remote.  In fact, they could easily repurpose a Bluetooth remote that connects to your phone, which then relays commands to the drone.  I envision something not much different than Garmin or GoPro’s remotes:

image

Ideally, this would have one core function: Changing view modes.

There’s not even a need for a start/stop button, since it automatically records video when it takes off.

I’d like to be able to have a few predefined or customizable buttons that automatically put it in follow or lead mode, or raise or lower the altitude a preset amount.  Also, the ability to tell it to just ‘catch up’, and by that I mean simply find a way to go up and get above the whole fray and catch up to me.

And since I’m on requests – I’d like it to be voice controlled. Something like this:

Me: ‘OK Skydio – change to lead mode’
Skydio: ‘Beep-boop’
Me: ‘OK Skydio – orbit me’
Skydio: ‘Beep-boop’

And so on.  Heck, even being able to set timers to say ‘Ok Skydio, in 10 seconds start an orbit’.

This may sound silly, but it’s these sorts of things that make the experience seamless.  If you’re mountain biking along as I was, it’s not exactly easy to try and tap the screen in exactly the right place on a hot and sweaty day.  Same goes for skiing, or any other sports.

Whereas a tiny pod controller that I could talk to or press quickly makes a world of difference.  After all, there’s a good reason why both Stalker and AirDog have controller pods – it makes it easy to quickly adjust the drone.

And ultimately – as fun as it is to get a single continuous shot of you going down the hill – the real future of this type of video is having it get the right shot of you going down that hill.  Knowing which side to be on for the most epic sunset silhouette pic, and when to swing out in front of you for the coolest lead shots.  And until it knows that automatically, at least let me do it easily manually.

Which, don’t get me wrong – the Skydio is still by far (massive leaps and bounds far) the best drone for sports out there today.  But, these little tweaks that are relatively trivial compared to what they’ve done already – would make it indispensable.

The Future:

DSC_2019

In many ways, I think the Skydio R1 isn’t so much about the R1, but rather, about what happens down the road (if that makes sense).  See, I view the R1 as a bit of a test product for the company. A teaser of sorts.  And I think they view it the same way.

Sure, it’s a fully functional product in incredible condition for an unknown startup that just launched their first drone.  It’s clear they’ve actually headed the advice to ‘ship when it’s ready’. Seriously, that part is impressive (and the $80M in VC funding likely helps there).

But I’m not sure that it’s impressive enough for the $2,500 they want you to pay.  In that sense, I almost feel like the $2,500 is an artificial price barrier to limit production to lesser unit quantities while they work out whatever comes down the road.  Certainly, they’d have to know that releasing a $2,500 drone that only does sports follow-me is mostly a non-starter, unless it’s perfect (which it’s not). In the drone landscape, that price is dangerously close to DJI’s prosumer Inspire series (which doesn’t much do sports tracking).

And if I put on my non-accredited BOM (bill of materials) hat, I don’t see the components being anywhere near that price point, especially because they aren’t giving away margin to retailers as they sell only through their site.  Which to me means they’ve gotta be eyeing a price drop down the road.  The question becomes when, and to what level?

Let’s pretend $1,500: This would roughly be the same level that AirDog is and has been priced at. And while that might be successful in niche startup terms, it’s hardly a blockbuster seller by any means.  Simply put – $1,500 is still too much for what most people on YouTube want to spend on a drone, which is yardstick that matters here.

So then what about $999? Yup, that’s the magic money amount. Undoubtedly Skydio would want that value to be higher, but certainly they have to know they’re being compared to the DJI Mavic series, which floats in that $1,000 range.  Sure, different capabilities, but just like deciding on a new bike – if you have an allotment of only X amount for said bike, there are many ways you could spend that: Road, mountain, etc… All of which are very different, but it’s still just one wallets worth of money.

Of course – all of that ignores the elephant in the room: Acquisition.

That’s ultimately why there’s $80M in VC funding. Those investors are hoping for a big payday. They’re hoping that somebody comes along and buys out the company or its tech, and they can walk away richer.  But that’s a dangerous assumption to make, since realistically only one company these days has the means to do that: DJI. GoPro certainly doesn’t have the means or the interest to do that (especially after cancelling their drone aspirations earlier this year).  And the ugly reality is nobody else in the world cares or has the cash to care.

And $80M as a floor (which would undoubtedly be multiples higher) buys a lot of engineers for one thing: Obstacle avoidance and tracking.  DJI already knows the drone flying aspects, and is realistically the best in the business for those aspects.  So could DJI build out their own obstacle avoidance for under $80M? Of course they could.

Thus, in my mind the only viable path forward for Skydio is to double-down.  By that I mean: Drop the price and own the kick-ass product they’ve made, increase their retail distribution channels so normal consumers actually know about them, find a way to cover the gaps for non-sports use…and then they’d have a legit winner on their hands by summer or fall (even despite existing hardware form factor which some of the limitations like lack of folding).

Summary:

DSC_2037

There’s simply no better sports tracking drone in the market today. Period. The Skydio R1 does what countless drones before it have tried to do, and failed in one manner or another. It tracks you without crashing to the ground along the way, whereby even an occasional tree or branch bonk isn’t a flight ender.  Not to mention the 4K quality performed very well for me on all my flights.  Or the fact that they even include two batteries in every kit.

But in some ways this isn’t entirely about whether the unit is the best sports tracker in the market.  No, it’s about whether it’s viable at the price point currently set – $2,500.  And in that sense, I think it falls short. Outside of early geeks like myself, the market at that price point is simply unsustainable. And even myself, I can’t justify $2,500 for it.  But I can justify $1,000 or maybe $1,500 with some quirks addressed.  And if that were the case, I’d be first in line with my credit card.

It may sound odd, but it’s not very often I’m sad to send back loaner devices. Perhaps because I have so many devices already, or perhaps because whatever I’m sending back isn’t terribly different than what I have.  But in this case when I hand it off to FedEx in the next day or so, I’ll be a bit sad to not see where development takes it.  Certainly my discussions with Skydio have them hinting at ways they can update the firmware to address some of the concerns I have (such as speed limitations or cautiousness), and I’ve also gotta believe they’ll find ways to add and tune some of the cinematic aspects as well – all of which will undoubtedly make it a more impressive product.

After all, in software speak – R1 simply stands for ‘Release 1’, which by its very definition implies a future numerical collection of drones.  With that I look forward to seeing where the company can take the R1, not just in price, but also features.  As well as where they can take the technology into other products.

Thanks for reading!

Source: dcrainmaker

Leave a Reply

Your email address will not be published. Required fields are marked *