Fixed Wing Drone suggestions for DSAR work?

Johnny C! · Jul 7, 2019

And to clarify my the other side of the equation.

Does a fixed wing drone have a practical use on DSAR?

The ability to fly it out of VLOS is obvious, so is this
range capability lost due to our VLOS restrictions?
And if so, I still can't help but think that the loiter time
can be be very useful.

Input please.

Thanks!

John

clolsonus · Jul 7, 2019

I would focus on thinking through your workflow and worry less about the specs. Is landing and changing a battery every 15-20 minutes ok? Do you have the time/experience(desire) to DIY something from parts? You will need to maintain your DIY system yourself (and it will take care and feeding.) Will you focus more on real time video, or more on assembling still shots into larger maps after the flight? Do you anticipate being able to find open areas with clear approach paths for landing, or will you be stuck operating from tight areas with lots of trees and other obstructions? Are you flying fairly high to get a broad overview of the area, or flying very low to get as much pixel detail as possible?

For my work (invasive plant species detection) the answers ended up being both yes and no for all of those. So we bought an off-the-shelf DJI phantom 4 and DIY'd our own fixed wing solution. Some days and some areas are better for one airframe and some situations for the other.

I will comment that on the whole, I like our fixed wing solution better than the phantom most days. But we have never put a scratch on our phantom yet in 20+ flights. In a similar number of flights, our fixed wing aircraft has seen some rough landings ... one in a frozen/rough plowed field, one bounced off the top off a cut corn stalk that spun it around. So our fixed wing, even though it's more capable in many ways, is also more work to operate and has taken some patching. Also I wrote the entire autopilot and flight control software stack from scratch (opensource on github), so if anything glitches, it's on me to fix. I really enjoy that process, and I'm able to achieve some goals that would be hard to work out on other popular DIY flight control systems, but maybe not too many other people in the world want to mess with the system at this level?

There never is one perfect or right solution for everyone, so at some point, you just have to make your best call, pull the trigger, buy some stuff, put it together, and go fly. You will quickly gain experience and perspective as you move forward. Expect a v2.0 (probably sooner rather than later) and work to improve each new system in some substantial way over the previous. Don't feel like you need to source (or build) the perfect system on the first try, and realistically plan on some attrition with your equipment.

BigAl07 · Jul 7, 2019

John I think the VLOS limitation is going to be a big show stopper for many long-range systems. With that being said, I have been thinking about this fixed wing system as a possible experimental platform:

R Martin · Jul 8, 2019

Johnny C! said:
@R Martin,

The FireFly 6 Pro is going to be out of my price
range at this time, but I like the Vtol/fixed wing
package. It combines the best of both types of
drones.

Also, in a effort to help me better understand the
possible advantages of a fixed wing drone, please
describe your current mission ops.

Thanks for the input!

John

We are currently using it for mapping large areas/construction sites to document property/construction changes. I am using the Sony RX1R II camera to do the work from an altitude of 330 ft AGL on an automated flight plan to ensure coverage of the site of interest. So far we are noticing a linear range of approximately 18-19 miles and a flight time of roughly 30 minutes on one battery pair which covers approximately 120 acres of ground. We also utilize the aircraft's terrain following mode to maintain an exact height above ground (which translates to a precise 0.5 in/px GSD). We have noticed that the onboard GPS unit is accurate to about 0.75 FT without the base station providing an RTK correction stream to the aircraft. Through repeated flight testing the aircraft hit the 4 foot target upon landing 100% of the time without operator intervention (for point of reference our DJI lands within 15 feet of the target).

The strength of the aircraft is much larger area coverage and more endurance in less time. The aircraft has a reliable GPS unit out of the box without the base station that, with proper ground control, can accurately map a job site. I suspect that the addition of the GPS base station to the mix will negate the need for GCPs (though they will instead become check points for data validation). We own three battery pairs though charging in the field takes approximately 35-45 minutes. Post processing is a little more cumbersome due to the fact that the RX1R II does not write geoinformation to the EXIF data. That requires an additional step prior to processing the imagery (not a big deal but it certainly adds to the processing time).

The limitations that we have found to date is that the aircraft has not (as of this moment) met the 55 minute flight time per battery pair( we are getting about 30 minutes in 85-94 degree weather which may be a contributing factor). The aircraft does not perform well in manual flight in hover mode (granted, it is not designed to). We operate a custom built graphics workstation that we use exclusively for imagery processing. My last job with ~1100 images took upwards of 38 hours to process. 42MP images are approximately 4-5 megabytes a piece. This has prompted us to buy another graphics workstation to keep from falling behind. Another limitation for this particular setup is there is no real-time imagery available during flight on the base station. That might be possible with an additional laptop and a gimbaled setup (I have no reason to use this at this time). The aircraft requires a laptop to run as a base station. Tablets do not work. The GCS software needs to be running and you must have the telemetry antenna connected in order to communicate with the aircraft during flight. You can also run the RTK GPS antenna as well (we bought a Toughbook due to the environment we operate in).

So the cost of the unit and the additional peripherals to support flight operations is considerably more than a prosumer choice but for our application, the FireFLY 6 Pro was an excellent choice and is performing its role very well.

R Martin · Jul 8, 2019

Johnny C! said:
I was hoping to gain insight into an affordable
air frame that I could utilize for DSAR, but I
don't know enough to answer the flight spec
question. As for the budget, I have been close
to spending $2k for an outfitted M2P mainly
for still images for commercial work.. If I find
a fixed wing platform conducive for DSAR in
that price range I would be closer to buying one,
but I will need to have a some previously
mentioned confabs with the experts here to
confirm that a fixed wing drone could be useful
for DSAR.

I guess the BLUF is, I don't know what I don't
know, and that is why I am here.

$2K will not touch a fixed-wing that does what you need it to do. $2K won't touch most of the sensor packages. Bare bones you can probably get in the ballpark with $20K though you are still going to want to add additional equipment to support flight operations.

clolsonus · Jul 8, 2019

I've had some jobs in areas of significant terrain and have started to wonder about terrain following strategies. I'm curious what terrain data source(s) the various systems use? I'm familiar with SRTM, but it has at least 5-10m vertical error and doesn't account for trees or other obstructions. Does the firefly do anything fancy if the terrain below it begins to rise faster than it can climb? Do any of the flight planning software tools do anything fancy like creating routes that follow the contours of the terrain (to minimize climbs and descents)? I love hearing the details of people's practical experiences with their tools.
For my work (so far) I've managed to find a reasonable high spot in my survey area, launch/land from there, and then fly typical grid survey routes ... but every new site I go visit has new and unexpected challenges. The open field in google earth now has 1000's of seedlings planted in it, didn't see the power lines in the map, the terrain ends up being way steeper than it looked online, trees end up being way taller and way closer to the open area, etc.

... life can be especially challenging when operating fixed wing in new areas.

R Martin · Jul 8, 2019

clolsonus said:
I've had some jobs in areas of significant terrain and have started to wonder about terrain following strategies. I'm curious what terrain data source(s) the various systems use? I'm familiar with SRTM, but it has at least 5-10m vertical error and doesn't account for trees or other obstructions. Does the firefly do anything fancy if the terrain below it begins to rise faster than it can climb? Do any of the flight planning software tools do anything fancy like creating routes that follow the contours of the terrain (to minimize climbs and descents)? I love hearing the details of people's practical experiences with their tools.
For my work (so far) I've managed to find a reasonable high spot in my survey area, launch/land from there, and then fly typical grid survey routes ... but every new site I go visit has new and unexpected challenges. The open field in google earth now has 1000's of seedlings planted in it, didn't see the power lines in the map, the terrain ends up being way steeper than it looked online, trees end up being way taller and way closer to the open area, etc. ... life can be especially challenging when operating fixed wing in new areas.

SRTM does have inaccuracies built in but we are talking about flying at altitude; not nap of earth. Living in an area that is relatively flat compared to the East Coast or Rockie Mountain region, SRTM has worked flawlessly. The largest deviation to date we have faced is a 55 foot terrain variation which the flight planning software accounted for without any problem. We do have quite a few high voltage transmission lines in the area but my programmed flight altitude allows for them with plenty of clearance. As far as contour following, that is not one of the abilities of the FireFLY. You do have to program a base altitude prior to creating the flight plan and that is based upon GSD AND terrain clearance.

clolsonus · Jul 8, 2019

R Martin said:
SRTM does have inaccuracies built in but we are talking about flying at altitude; not nap of earth. Living in an area that is relatively flat compared to the East Coast or Rockie Mountain region, SRTM has worked flawlessly. The largest deviation to date we have faced is a 55 foot terrain variation which the flight planning software accounted for without any problem. We do have quite a few high voltage transmission lines in the area but my programmed flight altitude allows for them with plenty of clearance. As far as contour following, that is not one of the abilities of the FireFLY. You do have to program a base altitude prior to creating the flight plan and that is based upon GSD AND terrain clearance.

I'm in minnesota so you would expect things to be pretty flat, but down along the Mississippi river there are areas with extremely steep wooded terrain with elevation changes of 300-400' from the bottom of the valley to the top of the bluffs. It's not the colorado rockies, but just as steep for that 300' elevation change, so pretty challenging for small UAV operations. That's where one of my projects takes me.

In my DIY system, the aircraft itself computes the actual survey route on board. At the ground station you just draw a polygon outline and send it up. That way you don't have to wait for 400-1000 waypoints to (maybe?) get up to the aircraft successfully before launch, or you don't need a ground station app trickling the waypoints up as the flight progresses (and maybe?) doing that successfully every time. I also like that the airplane can generate the route to account for actual measured winds aloft. I like to fly my transacts perpendicular to the wind, and then do my 180 turn upwind at the ends ... that keeps the turn radius as tight as possible. It would be quite an interesting programming challenge to come up with an algorithm to generate terrain contour following transacts that balance wind direction and somehow produce a route that maximizes efficiency, while maintaining safe terrain clearance at all times. Even more challenging with fixed wing because you can't stop and climb straight up if the slope gets really steep. Hmmm ...

Johnny C! · Jul 8, 2019

R Martin,

Thanks for the detailed write up.

You mention that your flight distances are 19-20 miles.
Are you able to maintain VLOS during these operations?

Thanks again!

John

R Martin · Jul 8, 2019

clolsonus said:
I'm in minnesota so you would expect things to be pretty flat, but down along the Mississippi river there are areas with extremely steep wooded terrain with elevation changes of 300-400' from the bottom of the valley to the top of the bluffs. It's not the colorado rockies, but just as steep for that 300' elevation change, so pretty challenging for small UAV operations. That's where one of my projects takes me.

In my DIY system, the aircraft itself computes the actual survey route on board. At the ground station you just draw a polygon outline and send it up. That way you don't have to wait for 400-1000 waypoints to (maybe?) get up to the aircraft successfully before launch, or you don't need a ground station app trickling the waypoints up as the flight progresses (and maybe?) doing that successfully every time. I also like that the airplane can generate the route to account for actual measured winds aloft. I like to fly my transacts perpendicular to the wind, and then do my 180 turn upwind at the ends ... that keeps the turn radius as tight as possible. It would be quite an interesting programming challenge to come up with an algorithm to generate terrain contour following transacts that balance wind direction and somehow produce a route that maximizes efficiency, while maintaining safe terrain clearance at all times. Even more challenging with fixed wing because you can't stop and climb straight up if the slope gets really steep. Hmmm ...

That is where you break things up into smaller blocks of airspace to account for drastic changes in elevation.

R Martin · Jul 8, 2019

Johnny C! said:
R Martin,

Thanks for the detailed write up.

You mention that your flight distances are 19-20 miles.
Are you able to maintain VLOS during these operations?

Thanks again!

John

It gets challenging and sometimes on the back side of the box I lose LOS but I also identify those areas prior to flight on my site walk and station VO's to cover me. We are all tied together with VHF sets and good communication procedures cover gaps in momentary LOS loss. And just to be clear before anyone gets their skivvies in a bunch, 19-20 LINEAR miles of flight; not 19-20 away from the base station.

Johnny C! · Jul 8, 2019

R Martin,
I thought that it would be a linear distance,
but that's still a big area for VLOS.

Thanks again for the input.

John

R Martin said:
It gets challenging and sometimes on the back side of the box I lose LOS but I also identify those areas prior to flight on my site walk and station VO's to cover me. We are all tied together with VHF sets and good communication procedures cover gaps in momentary LOS loss. And just to be clear before anyone gets their skivvies in a bunch, 19-20 LINEAR miles of flight; not 19-20 away from the base station.