In the hopes that this helps others, I just want to post my experiences with my missing in action P2V+. Much thanks to those on the thread with useful tips. (http://www.phantompilots.com/viewtopic.php?t=32351). Chalk up one more phantom pilot experience suggesting that you always fly your drone with a tracker. This really should be mandatory for anyone who is serious about being able to recover their UAV. At thus time I do not have my drone, and at this point I've had to conclude it has either vanished or been stolen. Without a second drone to perform a search, I cannot cover the remaining un-searched places (flat roofs and a nasty freezing swamp with poor visibility.)
What happened
I was on day 5 of my new Phantom, flying in 32 degree weather on New Years Day, over at Danehy Park in Cambridge, MA. Wind was relatively calm (10mph gusts). The field is open, relatively unpopulated in the winter, and has few trees. I conducted a number of successful test flights with one battery, including a very lengthy 2000 foot autonomous flight around the park. I then switched to my second battery, and after some close-to-ground maneuvers, took off for a longer FPV flight, intending to complete a circle after heading around the eastern border of the park.
The crash
At a distance of 800 feet, and a height of 392 feet, I received my last video transmission from the phantom. Around that time, the wind picked up considerably. Pretty much as soon as the WIFI clicked off, I activated return to home, first using S1, and then later by powering the transmitter off altogether. There should have been 5 minutes of battery left which would have been several times over more than enough time to return to the home location. The drone was slightly beyond visual range, hidden by the cusp of the hill. I waited 5 painful minutes to see if there was any sign of the drone but it never came back. In retrospect, I should have started moving towards the drone to re-establish visual contact but I was worried that my unpacked flight gear might get stolen and in any case felt fairly certain the drone must be on its way back.
Last recorded image:
Attempted recovery
My first attempts to recover the drone were frustrated by some contradictory information from the drone that I still don't understand. The last reported GPS location showed the drone clearly in the bounds of the park, in a swampy wetland area. My first steps were to go immediately to the area the drone was last reported by GPS, a wetland area at the edge of the park covered in ice. Not seeing it there, or anywhere around the flight path, I concluded that it must have been blown further east. I then scoured the surrounding neighborhoods, as well as the back area of the park. In the process I found a number of lost kites, and even a lost model rocket, but no sign of the drone. There are no leaves on the trees here in New England, so mostly no foliage to hide.
I began to suspect that the GPS coordinates were (somehow?) wrong. The frustrating part: the last visual does not correspond to the GPS coordinates. Does the GPS telemetry occur linked to the video? Or if the video is lost, do the coordinates still transmit with the available bandwidth? Its hard to tell. If they were sending data with UDP or multiple TCP streams, it is feasible that the GPS could continue to refresh even if the video was lost. Back at home I reconstructed the location of the last video transmission in google earth:
This seems to show the drone was actually outside of the bounds of the park, at least 100-200 feet to the east of the reported GPS position.
Of course, the Phantom's camera is fairly wide angle, and google earth's rendering is not, so I suspect that in reality the drone was even further south, perhaps even to the bend in the road nearby, than what is shown in the screen shot. So now comes some guesswork. Say that that the drone made it to the white roofed building, and then started trying to come back when it crashed. That would explain why its last transmitted GPS location was closer to the home location than the video.
In any case, I have already covered that area on foot, checking trees, roofs, back yards. The drone almost certainly came down in what I estimate was a 500 foot radius around its last reported location. I have been over that area with a buddy now multiple times.
In desperation, i went back over to the GPS coordinates a second time with a buddy to spot me, and sacrificing a good pair of shoes, waded into the freezing cold swamp to verify the drone wasn't hiding in the reeds. The entire swamp area "floor" (e.g. plane of the water) is almost impossible to see, even from the surrounding hills, because dead leaves and reeds make a large "brambly nest" 2-3 feet high. The irony: Searching for the drone would be much easier if I had another UAV to perform a visual "from the air" search. I tried to follow the tips in the sticky thread on using the wifi signal to locate the drone.
A frustrating 48 hour search
Returning later that day and the next day, I canvassed the entire area, placing posters on telephone poles, people's car windsheilds, and mailboxes. I distributed over 75 posters, and covered almost 8 miles on foot (according to my fitbit!) Protip: Searching for a lost drone is really great exercise. I also posted on craigslist. At this point, posters have been up for more than 48 hours, and no calls or responses. Its a fairly dense suburban area in Boston, and that neighborhood seems to have a lot of fairly nice people with families based on the folks I ran into during the search. My honest gut feeling is that if anyone had found it and even been remotely willing to return it, they would have by now. I suppose there is a small chance that it fell into a really non-obvious place like a dumpster, a deck, or possibly on a flat roof. Without a second drone to conduct the search, its probably impossible to find out where it is.
What went wrong?
I think the issue is that in the cold, the battery life estimate actually was quite incorrect and the drone battery failed shortly after loosing signal. I wasn't aware the copter was fighting to maintain its position in the wind when it started gusting badly, and perhaps the drone tried unsuccessfully to return to home. I know for a fact that the home location was properly set because I test this on every battery change. In any case, this experience is probably at least 90% operator error and bad luck.
Things I will do differently in the future:
Things I am glad I did:
I would love a modification that indicates how much "effort" the drone is spending to maintain its location. Part of the problem is that DJI has made these drones very easy to fly, and extremely stable with the GPS. That means that a drone could be fighting to maintain position in wind and the operator can be on the ground blithly unaware. Also, the experience with the downed WIFI tells me that whatever tracker I buy or make really ought to have its own independent battery system.
Thanks for listening!
What happened
I was on day 5 of my new Phantom, flying in 32 degree weather on New Years Day, over at Danehy Park in Cambridge, MA. Wind was relatively calm (10mph gusts). The field is open, relatively unpopulated in the winter, and has few trees. I conducted a number of successful test flights with one battery, including a very lengthy 2000 foot autonomous flight around the park. I then switched to my second battery, and after some close-to-ground maneuvers, took off for a longer FPV flight, intending to complete a circle after heading around the eastern border of the park.
The crash
At a distance of 800 feet, and a height of 392 feet, I received my last video transmission from the phantom. Around that time, the wind picked up considerably. Pretty much as soon as the WIFI clicked off, I activated return to home, first using S1, and then later by powering the transmitter off altogether. There should have been 5 minutes of battery left which would have been several times over more than enough time to return to the home location. The drone was slightly beyond visual range, hidden by the cusp of the hill. I waited 5 painful minutes to see if there was any sign of the drone but it never came back. In retrospect, I should have started moving towards the drone to re-establish visual contact but I was worried that my unpacked flight gear might get stolen and in any case felt fairly certain the drone must be on its way back.
Last recorded image:
Attempted recovery
My first attempts to recover the drone were frustrated by some contradictory information from the drone that I still don't understand. The last reported GPS location showed the drone clearly in the bounds of the park, in a swampy wetland area. My first steps were to go immediately to the area the drone was last reported by GPS, a wetland area at the edge of the park covered in ice. Not seeing it there, or anywhere around the flight path, I concluded that it must have been blown further east. I then scoured the surrounding neighborhoods, as well as the back area of the park. In the process I found a number of lost kites, and even a lost model rocket, but no sign of the drone. There are no leaves on the trees here in New England, so mostly no foliage to hide.
I began to suspect that the GPS coordinates were (somehow?) wrong. The frustrating part: the last visual does not correspond to the GPS coordinates. Does the GPS telemetry occur linked to the video? Or if the video is lost, do the coordinates still transmit with the available bandwidth? Its hard to tell. If they were sending data with UDP or multiple TCP streams, it is feasible that the GPS could continue to refresh even if the video was lost. Back at home I reconstructed the location of the last video transmission in google earth:
This seems to show the drone was actually outside of the bounds of the park, at least 100-200 feet to the east of the reported GPS position.
Of course, the Phantom's camera is fairly wide angle, and google earth's rendering is not, so I suspect that in reality the drone was even further south, perhaps even to the bend in the road nearby, than what is shown in the screen shot. So now comes some guesswork. Say that that the drone made it to the white roofed building, and then started trying to come back when it crashed. That would explain why its last transmitted GPS location was closer to the home location than the video.
In any case, I have already covered that area on foot, checking trees, roofs, back yards. The drone almost certainly came down in what I estimate was a 500 foot radius around its last reported location. I have been over that area with a buddy now multiple times.
In desperation, i went back over to the GPS coordinates a second time with a buddy to spot me, and sacrificing a good pair of shoes, waded into the freezing cold swamp to verify the drone wasn't hiding in the reeds. The entire swamp area "floor" (e.g. plane of the water) is almost impossible to see, even from the surrounding hills, because dead leaves and reeds make a large "brambly nest" 2-3 feet high. The irony: Searching for the drone would be much easier if I had another UAV to perform a visual "from the air" search. I tried to follow the tips in the sticky thread on using the wifi signal to locate the drone.
A frustrating 48 hour search
Returning later that day and the next day, I canvassed the entire area, placing posters on telephone poles, people's car windsheilds, and mailboxes. I distributed over 75 posters, and covered almost 8 miles on foot (according to my fitbit!) Protip: Searching for a lost drone is really great exercise. I also posted on craigslist. At this point, posters have been up for more than 48 hours, and no calls or responses. Its a fairly dense suburban area in Boston, and that neighborhood seems to have a lot of fairly nice people with families based on the folks I ran into during the search. My honest gut feeling is that if anyone had found it and even been remotely willing to return it, they would have by now. I suppose there is a small chance that it fell into a really non-obvious place like a dumpster, a deck, or possibly on a flat roof. Without a second drone to conduct the search, its probably impossible to find out where it is.
What went wrong?
I think the issue is that in the cold, the battery life estimate actually was quite incorrect and the drone battery failed shortly after loosing signal. I wasn't aware the copter was fighting to maintain its position in the wind when it started gusting badly, and perhaps the drone tried unsuccessfully to return to home. I know for a fact that the home location was properly set because I test this on every battery change. In any case, this experience is probably at least 90% operator error and bad luck.
Things I will do differently in the future:
- Always have a spotter with me
- Run immediately to the last known location of the drone to try to establish contact using WIFI
- Make sure my name and contact info is clearly written on the drone
- No beyond visual range flight unless I have a cheap transponder or tracker (this could have easily saved my drone since the main issue was locating it)
- Land if wind conditions change
Things I am glad I did:
- Verified how RTH works - I knew very clearly how the drone would respond on loss of contact.
- After loosing video and visual contact, switching to RTH ASAP
- Kept screen shots of the drone's last reported information for later reconstruction
I would love a modification that indicates how much "effort" the drone is spending to maintain its location. Part of the problem is that DJI has made these drones very easy to fly, and extremely stable with the GPS. That means that a drone could be fighting to maintain position in wind and the operator can be on the ground blithly unaware. Also, the experience with the downed WIFI tells me that whatever tracker I buy or make really ought to have its own independent battery system.
Thanks for listening!
