Drone outlook: The next 5 years

Gary--

Full disclosure, I left BU to go work at CyPhy Works. My comments might be biased. ;)

What I love about the hobbyist community is that it does an invaluable job of testing software and user interfaces. I know we did a lot of this together at OpenPilot and we've continued doing it after having moved on. But while fast innovation is critical, so is the plodding crossing of t's and dotting of i's. Developing an autonomous robust system that is capable of working in an industrial environment with >99% availability is extremely challenging. It's all those dreaded edge cases. Here's a quick list of things that pop into mind:

Our standards require immunity to a 200V/m EM field. That means that if you have an unshielded trace 3cm long, you will see 6V signals running up and down it. That will scramble pretty much every non-differential bus on the autopilot. This can be shielded against, but you pay a price in weight and payload--not counting the real money paid to the EM specialist to fix problems. For comparison, automotive standards are "only" ~30V/m.
Our wind requirements are rigorous. I don't have them in front of me, but I want to say we have to operate normally in >30kts of continuous wind at 130m AGL, and this while carrying full payload at 10,000' density altitude*. Beyond the controls challenge, there's a serious power penalty that must be addressed.
Temperature. One nice thing about testing in Massachusetts is the long, cold winter. It's surprising how many things will go wrong in these conditions. For instance, the venerable MPU-6000 accelerometer shows a >20% transient hysteresis when you cool it from above vs. when you cool it from below.
Another outcome from cold weather testing (and thank god there are some positives about it, because I've been testing this winter in -12C weather and it's no fun!) is that you can't rely on foam or other similar substances to isolate your IMU from vibration, since at these temperatures their hardness changes.
Servicing the industrial market is very different from that for hobbyists and researchers. A work-truck looks beat up after a year in service and our drones have to survive these same people. ;)

I optimistically think there are only a few more years of work to be done before we see large actors feel comfortable in depending on UAVs. Like Helen Greiner said in her talk, our demos are boring things, nothing like the thrilling videos on the internet. It's like watching an elevator go up and then come back down. It's taken a lot of unsexy work to get that far, but I don't know how else it could have been done. (If anyone has ideas, we're hiring!) Commercial aviation has set an incredibly high bar for safety and large-scale UAV operators will be expected to exceed that.

* P.S.: I love aviation. It's the only field where you can mix three different measurement systems and not look schizophrenic.

You need to be a member of diydrones to add comments!

Join diydrones

Comments

Greg Dronsky April 10, 2014 at 8:24am

heh wifes get use to it, just show her live view from fpv, and she will be amazed :)
Eric Stewart April 10, 2014 at 8:08am

^^^ LOL...

I build racecars and race them (naturally...)... I may be selling one to support my new hobby! If I tell my wife i'm buying a multi-rotor or fixed wing craft, she'll have a heart attack!
Greg Dronsky April 10, 2014 at 7:59am

@Eric Steward I am sure a man with a financial background doesn't have to be a great engineer to greatly support the private drone industry :)
Eric Stewart April 10, 2014 at 7:55am

I am new to this world, but I found the presentation by Helen Greiner to be highly inspirational. I work in the financial arena. I went straight to iRobot's financials to read about their investments in the UAS space. that's when I realized that CyPhy Works was a separate entity! lol... :(

I wish I were an engineer. All I've ever built are race cars...

Thank you for sharing!
Kenn Sebesta April 9, 2014 at 11:22pm
Gary--

Full disclosure, I left BU to go work at CyPhy Works. My comments might be biased. ;)

What I love about the hobbyist community is that it does an invaluable job of testing software and user interfaces. I know we did a lot of this together at OpenPilot and we've continued doing it after having moved on. But while fast innovation is critical, so is the plodding crossing of t's and dotting of i's. Developing an autonomous robust system that is capable of working in an industrial environment with >99% availability is extremely challenging. It's all those dreaded edge cases. Here's a quick list of things that pop into mind:
- Our standards require immunity to a 200V/m EM field. That means that if you have an unshielded trace 3cm long, you will see 6V signals running up and down it. That will scramble pretty much every non-differential bus on the autopilot. This can be shielded against, but you pay a price in weight and payload--not counting the real money paid to the EM specialist to fix problems. For comparison, automotive standards are "only" ~30V/m.
- Our wind requirements are rigorous. I don't have them in front of me, but I want to say we have to operate normally in >30kts of continuous wind at 130m AGL, and this while carrying full payload at 10,000' density altitude*. Beyond the controls challenge, there's a serious power penalty that must be addressed.
- Temperature. One nice thing about testing in Massachusetts is the long, cold winter. It's surprising how many things will go wrong in these conditions. For instance, the venerable MPU-6000 accelerometer shows a >20% transient hysteresis when you cool it from above vs. when you cool it from below.
- Another outcome from cold weather testing (and thank god there are some positives about it, because I've been testing this winter in -12C weather and it's no fun!) is that you can't rely on foam or other similar substances to isolate your IMU from vibration, since at these temperatures their hardness changes.
- Servicing the industrial market is very different from that for hobbyists and researchers. A work-truck looks beat up after a year in service and our drones have to survive these same people. ;)
I optimistically think there are only a few more years of work to be done before we see large actors feel comfortable in depending on UAVs. Like Helen Greiner said in her talk, our demos are boring things, nothing like the thrilling videos on the internet. It's like watching an elevator go up and then come back down. It's taken a lot of unsexy work to get that far, but I don't know how else it could have been done. (If anyone has ideas, we're hiring!) Commercial aviation has set an incredibly high bar for safety and large-scale UAV operators will be expected to exceed that.

* P.S.: I love aviation. It's the only field where you can mix three different measurement systems and not look schizophrenic.
Moderator

Gary Mortimer April 8, 2014 at 11:08pm

2009 called it want's its ideas back http://www.skysapience.com/ I can't believe that's really there thinking. What is not already being done right up to the 2018 bar?

I predict that OS hardware and software will be able to match or better their safety or stability claims by 2018. They have an in with the military so no doubt they will be able to sell their innovations there if they look all that exciting to the hobby tinkerer toy market time will tell.

The flying robot game is completely different to the ground robot world of 14 years ago. Way more players way more sexy. Innovation at a way faster rate.

This reply was deleted.