Google+ Flying a DJI Phantom 2 at Bardarbunga volcano in Iceland. About... - Drones HomeDrones Home

Flying a DJI Phantom 2 at Bardarbunga volcano in Iceland. About…

October 2, 2014 - Comment

Hosted at: http://skypixel.org/post/98868862799

[youtube http://www.youtube.com/watch?v=_L6Phuwqi7Y]

Here is a second video that just shows pretty aerial footage:

[youtube http://www.youtube.com/watch?v=8cICS9MtRRw]

The logistics of the operation were challenging. We had literally one day of suitable weather in the forecast to get close to the volcano. We needed a relatively clear day with low winds, but enough wind to push the poisonous gases away from our approach direction. With only 16 hours of notice, I boarded a flight to Reykjavik from San Francisco with 3 DJI Phantom 2 quadcopters in tow. I spent 12 hours in the plane, followed by an additional 15 hours in the car before we were on location at the Holuhraun eruption site.

Even with the proper permissions, we were stopped often by police and rangers while we were at the volcano. Each time, we showed our paperwork and were cleared. When we arrived, our vehicles were not allowed very close to the volcano (I estimate that we were about 2 km away). We flew anyway, using DJI Lightbridge for a live, high-definition video stream, and recorded the wireless footage locally using an Atomos Ninja Blade so we would be able to keep footage even if the Phantom went down for some reason. Unfortunately, we were too far away to fly into the caldera—we couldn’t fly all the way there without losing radio control to the Phantom (it would turn around and start coming home, each time). We spoke with the rangers about what we were trying to do, and they became really interested. Obviously, there is a lot a ranger might be able to do with Phantoms in the field, so we gave them a quick demo. Eventually, they told us that, although we couldn’t drive any closer, we could walk all the way to the lava flow. They were clear that this was not recommended, but that it wasn’t prohibited. So we walked down to the edge of the lava and set up to fly from there (about 1km away).

With a Phantom in the air at the caldera’s rim, I would periodically lose both video and control signal. There was definitely something about the eruption that was creating a lot of radio interference. I thought back to all the pictures I had seen of lightning above eruptions, which happens because of the static electricity generated by ash particles rubbing against each other. This particular eruption wasn’t an ash eruption, but something similar might have been going on. Luckily, DJI flight controllers tell their aircraft to return to home when they lose signal; we relied heavily on the autonomous return-to-home feature in order to retrieve our Phantoms at the end of each flight.

During the last flight, my video feed cut out completely while I was descending (in order to get closer to the giant, exploding pool of molten lava in the middle of the caldera). I flipped the Phantom into fail safe mode and waited for it to return to home. Sure enough, after a few minutes, the Phantom came flying back over us, and when it landed, we noticed that the front of the GoPro was melted! The camera no longer works, but luckily, the microSD card survived, and the footage is intact.

We walked back to the car and drove much further away from the volcano towards the river of lava. We were hoping to take pictures of the glow of the laval from afar, and perhaps be lucky enough to also capture northern lights (aurora borealis) in the same frame. Again, we were lucky, and I snapped a timelapse sequence of the northern lights dancing in the sky on top of the warm, red glow of a volcano erupting. How many times in one’s life should one expect to see such a thing? It was truly amazing.

HUGE thanks to Ragnar, Einar, Valdimar, Bryndis, and everyone else who helped to make this a reality. None of this would have been possible without you, and seeing the volcano erupting in front of us with northern lights dancing in the sky is a visceral memory I cherish forever!

Technical Notes

Capturing compelling footage with small-sensor cameras at night can be challenging. All of the aerial night footage in these videos were shot with a GoPro HERO 3+ Black Edition with Protune turned on, and EV compensation set to either -1.5 or -2.0. The correct exposure for this particular scenario would probably have been -1.75, but that isn’t a camera option. Shooting without EV compensation yields totally blown out bright things when they are on a dark background. This is common with GoPros and other cameras in any dark environment that has bright things on it (e.g., a white boat on the ocean’s surface, or lights at night).

I shot at 2.7K/30p instead of the more-cinematic 24p because I didn’t want to risk having jello in my shots, which can happen at 24p in dynamic environments. The daytime flights were shot at 1080/60p to minimize jello. Shooting at 24p or 30p is fine in calm conditions, but I had to fly at full speed toward the volcano for 2 minutes to reach it, and flying at full speed (in wind) can introduce jello into the video. Shooting at 60p can help to minimize jello, so I sometimes choose that frame rate when there is a lot of wind. I also shot 4K/15p in order to have high-resolution content from which I could extract still frames. The GoPro 4 has EV compensation in stills mode, but the 3 does not.

We used DJI Lightbridge for video and telemetry transmission only, opting to replace RC control in the Phantom with a 5.8G transmitter and receiver (an accessory DJI sells). We recorded locally on the ground in ProRes with an Atomos Ninja Blade so we could keep the footage in case the Phantom was lost. I turned OSD off so the video would be clean, and used a timer to time my flights. I could have used a second Lightbridge ground end with OSD turned on to monitor telemetry, but it was pitch black, and we had walked half a mile from the vehicles. We didn’t have enough hands, tripods, tape, or time to make that happen. :)

As I mentioned above, range was an issue from our first location, which was about 2km away from the caldera’s rim. When we moved closer, we were about 1km away, which should have been no problem for our radios, but we still lost both RC and video signal consistently when flying right above the lava. I suspect that the problem stemmed from RF interference from volcanic activity. The raw video is funny because the Phantom keeps turning around to try to come home each time I try to push it closer. You can see me constantly taking control and trying to force the aircraft back into the volcano. If we were to attribute autonomous thinking to the Phantom, we would have assumed that it was survival instinct!

Failsafe / return to home worked beautifully, and is the only reason we still have the Phantoms we used in Iceland. I’ve never had it not work in all of my time flying DJI flight controllers. Having said that, I am NOT recommending relying on the feature. You should be in full command control in the vast majority of situations!

Update: Wired just posted a story about it.

Hosted at: http://skypixel.org/post/98868862799

Here is a second video that just shows pretty aerial footage:

The logistics of the operation were challenging. We had literally one day of suitable weather in the forecast to get close to the volcano. We needed a relatively clear day with low winds, but enough wind to push the poisonous gases away from our approach direction. With only 16 hours of notice, I boarded a flight to Reykjavik from San Francisco with 3 DJI Phantom 2 quadcopters in tow. I spent 12 hours in the plane, followed by an additional 15 hours in the car before we were on location at the Holuhraun eruption site.

Even with the proper permissions, we were stopped often by police and rangers while we were at the volcano. Each time, we showed our paperwork and were cleared. When we arrived, our vehicles were not allowed very close to the volcano (I estimate that we were about 2 km away). We flew anyway, using DJI Lightbridge for a live, high-definition video stream, and recorded the wireless footage locally using an Atomos Ninja Blade so we would be able to keep footage even if the Phantom went down for some reason. Unfortunately, we were too far away to fly into the caldera—we couldn’t fly all the way there without losing radio control to the Phantom (it would turn around and start coming home, each time). We spoke with the rangers about what we were trying to do, and they became really interested. Obviously, there is a lot a ranger might be able to do with Phantoms in the field, so we gave them a quick demo. Eventually, they told us that, although we couldn’t drive any closer, we could walk all the way to the lava flow. They were clear that this was not recommended, but that it wasn’t prohibited. So we walked down to the edge of the lava and set up to fly from there (about 1km away).

With a Phantom in the air at the caldera’s rim, I would periodically lose both video and control signal. There was definitely something about the eruption that was creating a lot of radio interference. I thought back to all the pictures I had seen of lightning above eruptions, which happens because of the static electricity generated by ash particles rubbing against each other. This particular eruption wasn’t an ash eruption, but something similar might have been going on. Luckily, DJI flight controllers tell their aircraft to return to home when they lose signal; we relied heavily on the autonomous return-to-home feature in order to retrieve our Phantoms at the end of each flight.

During the last flight, my video feed cut out completely while I was descending (in order to get closer to the giant, exploding pool of molten lava in the middle of the caldera). I flipped the Phantom into fail safe mode and waited for it to return to home. Sure enough, after a few minutes, the Phantom came flying back over us, and when it landed, we noticed that the front of the GoPro was melted! The camera no longer works, but luckily, the microSD card survived, and the footage is intact.

We walked back to the car and drove much further away from the volcano towards the river of lava. We were hoping to take pictures of the glow of the laval from afar, and perhaps be lucky enough to also capture northern lights (aurora borealis) in the same frame. Again, we were lucky, and I snapped a timelapse sequence of the northern lights dancing in the sky on top of the warm, red glow of a volcano erupting. How many times in one’s life should one expect to see such a thing? It was truly amazing.

HUGE thanks to Ragnar, Einar, Valdimar, Bryndis, and everyone else who helped to make this a reality. None of this would have been possible without you, and seeing the volcano erupting in front of us with northern lights dancing in the sky is a visceral memory I cherish forever!

Technical Notes

Capturing compelling footage with small-sensor cameras at night can be challenging. All of the aerial night footage in these videos were shot with a GoPro HERO 3+ Black Edition with Protune turned on, and EV compensation set to either -1.5 or -2.0. The correct exposure for this particular scenario would probably have been -1.75, but that isn’t a camera option. Shooting without EV compensation yields totally blown out bright things when they are on a dark background. This is common with GoPros and other cameras in any dark environment that has bright things on it (e.g., a white boat on the ocean’s surface, or lights at night).

I shot at 2.7K/30p instead of the more-cinematic 24p because I didn’t want to risk having jello in my shots, which can happen at 24p in dynamic environments. The daytime flights were shot at 1080/60p to minimize jello. Shooting at 24p or 30p is fine in calm conditions, but I had to fly at full speed toward the volcano for 2 minutes to reach it, and flying at full speed (in wind) can introduce jello into the video. Shooting at 60p can help to minimize jello, so I sometimes choose that frame rate when there is a lot of wind. I also shot 4K/15p in order to have high-resolution content from which I could extract still frames. The GoPro 4 has EV compensation in stills mode, but the 3 does not.

We used DJI Lightbridge for video and telemetry transmission only, opting to replace RC control in the Phantom with a 5.8G transmitter and receiver (an accessory DJI sells). We recorded locally on the ground in ProRes with an Atomos Ninja Blade so we could keep the footage in case the Phantom was lost. I turned OSD off so the video would be clean, and used a timer to time my flights. I could have used a second Lightbridge ground end with OSD turned on to monitor telemetry, but it was pitch black, and we had walked half a mile from the vehicles. We didn’t have enough hands, tripods, tape, or time to make that happen. :)

As I mentioned above, range was an issue from our first location, which was about 2km away from the caldera’s rim. When we moved closer, we were about 1km away, which should have been no problem for our radios, but we still lost both RC and video signal consistently when flying right above the lava. I suspect that the problem stemmed from RF interference from volcanic activity. The raw video is funny because the Phantom keeps turning around to try to come home each time I try to push it closer. You can see me constantly taking control and trying to force the aircraft back into the volcano. If we were to attribute autonomous thinking to the Phantom, we would have assumed that it was survival instinct!

Failsafe / return to home worked beautifully, and is the only reason we still have the Phantoms we used in Iceland. I’ve never had it not work in all of my time flying DJI flight controllers. Having said that, I am NOT recommending relying on the feature. You should be in full command control in the vast majority of situations!

Update: Wired just posted a story about it.

Comments

Leave a Reply