Agribotix test pilots educate 5th graders about responsible drone use

 

Agribotix test pilots Tom McKinnon and Rick Burd spent a morning educating 5th graders on responsible drone use in precision agriculture. They demonstrated how to program a flight path, launch the drone, and analyze the images. Additionally, they all spent part of the morning learning about dry land organic farming from Golden Prairie owners Jean, Randy, and Bryce, riding in a tractor, and piloting an R/C aircraft. Agribotix graciously thanks Golden Prairie for organizing this opportunity and looks forward to our next occasion to spread the word about drones for good. Please don't hesitate to contact us if you think your organization could benefit from a collaboration or demonstration.

Tom McKinnon showing a pair of 5th graders how to input a mission

Tom McKinnon showing a pair of 5th graders how to input a mission

The Agribotix crew & co. admiring one of Golden Prairie's tractors.

The Agribotix crew & co. admiring one of Golden Prairie's tractors.


 

Tasman Glacier survey completed

 

After nine separate drone flights (not including the one that led to the loss of Sky Hunter), we successfully surveyed the lower portion of Tasman Glacier! The blue dots in the image below show the locations of all of our camera shots -- a total of 2,735 in-flight photos. To give you a sense of scale, the survey covered 6km lengthwise, and, on average, just over 2km in width.

Blue markers show aerial photo locations atop Tasman Glacier. The full survey spanned over 6km in length along the glacier.

Blue markers show aerial photo locations atop Tasman Glacier. The full survey spanned over 6km in length along the glacier.

Brian and Tom testing the control surfaces of the RV Jet, with our landing zone in the background.

Brian and Tom testing the control surfaces of the RV Jet, with our landing zone in the background.

Our final launch and landing location was further downglacier than we had gone before, and our landing sites were becoming increasingly more subpar. Luckily, New Zealand bush is both plentiful and generally soft, so we identified a hill slope that could serve as a padded landing zone, as can be seen in the image to the left. We're naming this drone landing technique the Anderson technique, after Brian landed the RV Jet in the hill slope five times with no damage to the plane.

Before our final day of flying, we returned to our flight logs to figure out what exactly had gone wrong with the Sky Hunter. While we first suspected that a strong wind had brought us down, it appears that the real culprit was our Sunnysky motor. Mission Planner showed that we were pulling zero or close to zero amps at the end of the failed Sky Hunter mission, and while the autopilot was asking for throttle, the plane was unable to deliver due to motor failure. After doing a second sweep of the moraine walls, we were also able to locate the Sky Hunter from above, where it is sitting exactly where we expected based on its communications with our ground station. Unfortunately, this still means that it would be too dangerous to retrieve, so we will leave New Zealand without it.

Despite this loss, we're pretty happy with how the full mission turned out -- and are looking forward to processing the images (using Agisoft PhotoScan Pro) to see how our DEM turned out. The glaciologists we were working with plan on returning next year to make an updated DEM to see how the glacier is changing. We're looking forward to staying in touch with them to hear about these results, as well as other uses of drones in New Zealand!

On that note, the Agribotix team remains in New Zealand for another week. We're done with our Tasman Glacier flying, and are moving to data processing, but would be happy to do more flying with our remaining drone. If there's anyone on the South Island following our progress who has an interesting project that could use some aerial photographs, contact us at info@agribotix.com ! 

Tom and Karen, part of the Agribotix team, are all smiles after a successful mission!

Tom and Karen, part of the Agribotix team, are all smiles after a successful mission!

 

 

 

 

RV Jet up!

 
The RV Jet in the practice field. It had already been through its fair share of crashes, so we knew it would be tough.

The RV Jet in the practice field. It had already been through its fair share of crashes, so we knew it would be tough.

After we had a beer to mourn the loss of our Sky Hunter, we re-grouped and pulled out Brian's drone, an RV Jet. We had previously tried to autopilot the RV Jet, but were having some trouble connecting it to the ground station, so we weren't sure if it would be possible to get it field-ready in time to complete our survey. Luckily, we were able to identify a single broken wire that was giving us trouble, and returned to our practice field for some test flights after fixing the wire.

Our first flights didn't seem promising -- the RV Jet seemed to only fly in circles of radiuses that would way too large for our glacier surveying domain, and we couldn't get it to follow an autopilot mission. To make matters worse, we had a bad crash into a river bank during landing, which led to a severely bent wing. Although we were all about to throw up our hands and give up on our Tasman Glacier surveying mission, we decided that it might be worth trying to fix the wing. Tools and tape came out of the cars, and we started gluing a new spar into the bent wing to straighten it out.

Wing repair time.

Wing repair time.

Brian and Tom at the ground control station, tuning up the RV Jet parameters. 

Brian and Tom at the ground control station, tuning up the RV Jet parameters. 

This in-field repair worked astonishingly well, and we returned to Mission Planner (our autopilot software) to tune the autopilot parameters. The RV Jet is quite an agile, fast plane, and we realized that a few parameters were more appropriate for a slower drone; hence, the wide turns that the drone would take regardless of our instructions. As we tuned up the parameters, flight skill increased considerably. Not, unfortunately, to the level of the Sky Hunter, but good enough to be relatively certain we would not inadvertently crash into a moraine wall. As the sun set and the temperature got considerably cooler, we tested its range, and determined that we could fly 3km away without losing RC control. At this point, we decided that we would return to Tasman Glacier tomorrow to finish the survey -- a much more optimistic outlook than we had had just a few hours ago!

The one remaining hitch is that we lost our primary camera with the Sky Hunter. Brian's camera was not triggering with an IR signal like it was supposed to, and our backup camera wouldn't load the Canon Hackers Development Kit that we use to trigger the camera every approx. 3 seconds or so. But after a bit more fiddling, we figured out the right buttons to push, and have the camera ready for action. Tomorrow, we'll head back to Tasman, and hope to reach the terminus with the RV Jet. 

 

Sky Hunter down

 
Karen putting together the tail of the Sky Hunter before its first flight of the day. 

Karen putting together the tail of the Sky Hunter before its first flight of the day. 

Our second day flying over Tasman Glacier was, unfortunately, not as successful as our first. We moved our ground control site a few kilometers downglacier, with the goal of finishing surveying the glacier with 3 more flights. The winds were calm near our site, so we felt confident that we would have a good flying day. Our first flight was largely successful, although we had to call the Sky Hunter home before it finished its last two transects across the glacier due to low battery. We suspected that our batteries were draining faster than usual because the plane was flying faster than its airspeed indicator suggested -- our groundspeed was 10m/s faster than our airspeed regardless of the direction we were flying, and it seemed unlikely that the winds were shifting direction at the same time that our plane was! After doing a quick airspeed sensor calibration, we were ready to send the Sky Hunter on its next mission.

After a beautiful launch, we sent the Sky Hunter diagonally across the glacier to the farthest point of the mission, which was almost 3km past our ground control station. The flight seemed smooth, although a strong gust of wind threw the Sky Hunter off-path in the middle of its first transect. Since the drone recovered without losing elevation, we continued on the mission. After flying two more transects across the glacier, the Sky Hunter was again thrown off its path, but this time it started rapidly descending. With autopilot staying on track but not staying at elevation, we took over the controls, and attempted to regain elevation. However, we had the additional problem that the Sky Hunter had been near the end of its transect, so steering it full speed ahead would have led to it crashing into a steep moraine wall. As we tried to steer away from the cliff and give the drone full throttle in an attempt to gain elevation, the drone instead started spiraling down towards the surface. Too soon, it became clear that we had crashed into the ground. 

Because the drone maintained telemetry, RC, and GPS signal during its entire flight (and crash), we could identify its crash location. Unfortunately, we believe that the drone is sitting at the base of a steep, unstable moraine wall, and at the top of an unstable scree slope. The location is at extremely high risk for a rock avalanche, and is one that even the best mountaineers would do their best to avoid. Thus, we decided that rescue attempts would be ill-advised, and went home for the day.

Between the two days of flying, we now have 1,655 geotagged images (like the one below!) that span around 6km of Tasman Glacier, which we are currently processing into a DEM. Despite this setback of the crash, we hope to be back in action with a different plane soon. We'll keep you updated!

One of the 1,655 images of Tasman Glacier that we've taken from the Sky Hunter. The glacier is covered in debris, hence its gray color. Blue melt ponds are on the surface.

One of the 1,655 images of Tasman Glacier that we've taken from the Sky Hunter. The glacier is covered in debris, hence its gray color. Blue melt ponds are on the surface.

 

100km of across-glacier flights completed!

 

We completed our first flights over Tasman Glacier today! Our launch and landing site was about an hour hike past the end of the road -- a walk made slightly more challenging by the fact that we were carrying all of our gear on our backs. But once we arrived, the weather was perfect: sunny, clear, and almost no wind. 

The beginning of our hike towards our field site. The Sky Hunter is packed up in the tall box on Brian's back.

The beginning of our hike towards our field site. The Sky Hunter is packed up in the tall box on Brian's back.

Brian setting up the ground station, and Tom adding the tail to the Sky Hunter.

Brian setting up the ground station, and Tom adding the tail to the Sky Hunter.

Tasman Glacier is approximately 2km across, with steep moraine walls on either side, meaning we had to be extra careful to make sure we didn't crash into the moraines at the edges of our domain. Since Mission Planner (our autopilot software) relies on google earth images as the basis for its maps, our fingers were crossed that the google earth topography would be accurate! But just to be sure, we stayed 200-300m away from the cliffs. 

Overall, the take-offs were pretty easy, although it's always a little scary seeing the drone make it's first forays over the glacier! Due to the steep and loose nature of the moraine walls, a drone lost on the glacier may become a drone lost forever, as foot travel on Tasman Glacier can be quite dangerous. Landings were slightly more challenging, because of the boulders that were strewn throughout our landing field, but all of them ended up being successful, and the Sky Hunter survived unscathed. 

Karen designing the next mission.

Karen designing the next mission.

In total, we did three flights from our ground station location, each spanning the 2km of Tasman Glacier and 1-2km along glacier, depending on how far we had to fly from our ground station to get to the beginning of our survey block for the flight. The flight transects were chosen to have 80% overlap between photos, ideal for construction of the DEMs. With all of these transects, the Sky Hunter successfully flew over 100km! We now have over 1000 photos with which to create a first DEM.


Tom deconstructing the Sky Hunter -- and showing off his 3D robotics kit.

Tom deconstructing the Sky Hunter -- and showing off his 3D robotics kit.

At the end of the day, we were all smiles, and the hike back down the mountain didn't feel quite so long! Tomorrow, we'll return to fly over the lower portion of the glacier -- including Tasman Lake.