Please select Into the mobile phone version | Continue to access the computer ver.
Flying indoors and obstacle avoidance
3100 14 2017-2-25
Uploading and Loding Picture ...(0/1)
o(^-^)o
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

I was flying last night inside a building in an area with plenty of open space and light and managed to crash into a flat wall. The mavic was flying under OPTI mode, using front and rear VPS sensors. Lighting conditions were bright so I don't see why the obstacle avoidance would not have detected the wall and slowed the drone down? I uploaded the flight data to healthy drones but it is not showing any data for notifications???
2017-2-25
Use props
Daroga
Captain
Flight distance : 3517287 ft
  • >>>
United States
Offline

Were you flying in Cinematic, tripod, or using any other IOC modes?
2017-2-25
Use props
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

Daroga Posted at 2017-2-25 11:11
Were you flying in Cinematic, tripod, or using any other IOC modes?

No, I was not flying in any of the specific flight modes. Turned on the drone, made sure VPS was active with the green double flash and the controller was showing OPTI mode.

I adjusted my camera settings and pressed record, the video file is corrupt and won't play, but I did not use any specific flight modes or go into any menus to adjust settings.

The obstacle avoidance was working earlier as I tested it by hovering in front of an object and when applying forward stick it would not move closer to the object.
2017-2-25
Use props
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

fansca9280ea Posted at 2017-2-25 16:29
Did you hit the wall head on? Why didn't you slow down/stop knowing there was a wall there?
I fly in my house all the time with all of the VPS turned off because of the annoying radar beeping. Indoors the common rule is. Don't push your right joystick up when there's a wall ahead.  Lol.. I'm sorry you crashed. Hopefully it was a very slow drift into the wall because of no gps lock and is only prop damage.

It was for the most part head on. It was a stupid mistake on my fault, I decided to fly not using the live video feed and just by watching the drone. I was flying from one end of the building going in a straight line and was going to make a turn around a pillar which was on the other end of the building. It wasn't like I said, there is a wall let's fly right into it. By the time I noticed that I was closing in on the wall and passed the pillar quicker than I expected it was too late. I decided to pull back on the right stick all the way hoping it would stop. Unfortunately I was too late with my reaction and the gimbal is broken, the rest of the drone is fine.

What I'm trying to figure out is why was the VPS not working and slowing the drone down before reaching the wall to bring it to a stop. Although I made a flying error, these high tech drones are suppose to avoid objects, especially those that are straight ahead and a flat wall. Am I missing something? or am I wrong in thinking this is an aircraft malfunction. I can understand how the drone might not detect a tree branch etc. but a flat wall? This drone should have stopped and even slowed its speed down before reaching the wall.

Yes, anyone can say this was a pilot error, which I'm not arguing. But for a drone that is sold with a VPS system that is designed to avoid obstacles and can't even stop for a flat surface is also a drone malfunction in my opinion.
2017-2-26
Use props
Jason Lane
Second Officer
Flight distance : 260144 ft
Philippines
Offline

Was the wall painted one color, with no discernible texture? I haven't tested the forward OA on a large plain surface such as a wall, and I'd never really wondered how well it would work, but now that you've mentioned this, I am wondering if the uniform texture of the wall simply gave the forward sensors nothing to "see" as an obstacle. The forward OA system relies solely on the two vision sensors. There are no ultrasonics on the front like there are on the bottom. Could it be that the Mavic simply isn't capable of recognizing a large, plain-colored wall? It would be very ironic if that were the case.

I suppose it's also possible that even though you said it was bright and there was enough light for the OA system to function, the light may have still been sub-optimal, and not quite bright enough for the system to work as well as it should.

BTW, you've used the term 'VPS' many times, but the Visual Positioning System refers more specifically to the downward sensors used to aid in positioning, and includes the bottom ultrasonic sensors. I don't believe the forwards sensors are considered part of the VPS.
2017-2-26
Use props
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

Jason Lane Posted at 2017-2-26 09:31
Was the wall painted one color, with no discernible texture? I haven't tested the forward OA on a large plain surface such as a wall, and I'd never really wondered how well it would work, but now that you've mentioned this, I am wondering if the uniform texture of the wall simply gave the forward sensors nothing to "see" as an obstacle. The forward OA system relies solely on the two vision sensors. There are no ultrasonics on the front like there are on the bottom. Could it be that the Mavic simply isn't capable of recognizing a large, plain-colored wall? It would be very ironic if that were the case.

I suppose it's also possible that even though you said it was bright and there was enough light for the OA system to function, the light may have still been sub-optimal, and not quite bright enough for the system to work as well as it should.

The wall has wall paper that comprises of two colors with large 3 feet stripes approximately, and has a texture within the color.

As for brightness, I was filming at ISO 400 with a shutter of 1/60th I believe. IMO it was bright in the area I was filming. If I was shooting at 1600 ISO I could understand, but at ISO 400 I would think there would be plenty of light for an obstacle avoidance sensor to work.

If these sensors are so finicky, they should limit the sensors to when they can work depending on light levels available to the camera and sensors. But the forward facing sensors were working as intended at the beginning of the flight as they were detecting objects and slowing the aircraft down.

If the system is just unreliable, then these drones shouldn't be advertised  as being capable of avoiding obstacles. Flying and knowing that the drone will detect objects to avoid crashes in comparison to the drone is equipped with sensors that may help in avoiding obstacles is a big difference and would change how most people fly their drones and the video shots they try to achieve.
2017-2-26
Use props
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

At the end of the day I have Care Refresh, so depending on repair cost I can just get the drone replaced but I would hate to use 1 of my 2 replacements and pay for it, when the drone has a system that is suppose to avoid a circumstance like that.
2017-2-26
Use props
Jason Lane
Second Officer
Flight distance : 260144 ft
Philippines
Offline

Ely324 Posted at 2017-2-26 10:05
The wall has wall paper that comprises of two colors with large 3 feet stripes approximately, and has a texture within the color.

As for brightness, I was filming at ISO 400 with a shutter of 1/60th I believe. IMO it was bright in the area I was filming. If I was shooting at 1600 ISO I could understand, but at ISO 400 I would think there would be plenty of light for an obstacle avoidance sensor to work.

That wall certainly doesn't sound like the untextured or uniform type of surface that DJI warns about, and it does sound like light should not have been an issue. Seems pretty nuts that the Mavic couldn't have seen it coming. Unless the Mavic's angle was a little more offset than you remember. eg. If it was drifting sideways a little just before impact. I'm not trying to lay any blame on you, just trying to hypothesize why the sensors didn't pick up that wall.
2017-2-26
Use props
hallmark007
Captain
Flight distance : 9812789 ft
  • >>>
Ireland
Offline

Ely324 Posted at 2017-2-26 10:12
At the end of the day I have Care Refresh, so depending on repair cost I can just get the drone replaced but I would hate to use 1 of my 2 replacements and pay for it, when the drone has a system that is suppose to avoid a circumstance like that.

Maybe post your logs up here, might help..
2017-2-26
Use props
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

Jason Lane Posted at 2017-2-26 10:31
That wall certainly doesn't sound like the untextured or uniform type of surface that DJI warns about, and it does sound like light should not have been an issue. Seems pretty nuts that the Mavic couldn't have seen it coming. Unless the Mavic's angle was a little more offset than you remember. eg. If it was drifting sideways a little just before impact. I'm not trying to lay any blame on you, just trying to hypothesize why the sensors didn't pick up that wall.

It was maybe a few degrees off of being exactly perpendicular, it was pretty head on. I think why the gimbal snapped at the mount point was due to the fact that when I pulled back on the right stick the front of the drone tilted up exposing the gimbal to impact. With not enough space to slow down it was still drifting in that forward motion and the front legs folded in first during impact and then allowing  the gimbal to hit against the wall. It almost looked like the mavic landed on the wall as it approached the wall with a steep tilt. If the gimbal cover was on, there possibly would have been no damage other than props, the body of the drone shows no damage other than some scraping/paint transfer to the cover of the battery from the props.
2017-2-26
Use props
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

hallmark007 Posted at 2017-2-26 10:48
Maybe post your logs up here, might help..

I uploaded my flight log to healthydrones but it didn't show much data at all. I couldn't see any of the normal information like notifications etc. that you would normally see.

How would I upload the data to here?
2017-2-26
Use props
hallmark007
Captain
Flight distance : 9812789 ft
  • >>>
Ireland
Offline

Ely324 Posted at 2017-2-26 15:51
I uploaded my flight log to healthydrones but it didn't show much data at all. I couldn't see any of the normal information like notifications etc. that you would normally see.

How would I upload the data to here?

You just need to post the link.
2017-2-26
Use props
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

hallmark007 Posted at 2017-2-26 15:53
You just need to post the link.

flight log
2017-2-26
Use props
hallmark007
Captain
Flight distance : 9812789 ft
  • >>>
Ireland
Offline


Thank you but unfortunately that won't tell us anything, hopefully you get sorted soon, maybe those cages prop guards are the answer for indoors.

Good luck..
2017-2-26
Use props
Ely324
lvl.4
Flight distance : 904888 ft
United States
Offline

Remarks:        User stick command error FLY060 1. Unit was in P-GPS mode and was responsive to RC input; 2. At t=630s, relative height=33m, unit recorded external impact, then rolled over and fell. 3. User didnt have any sattelites and was using the vision sensor to fly. Conclusion: User stick command error. Impact in air. No avoidance system in atti mode [Non warranty Repair]

I don't understand this, I was flying in opti mode and had vision sensors working as well as obstacle avoidance. How does it go from opti mode to atti mode all of a sudden?

And they say I was in PGPS as well. So was I in PGPS or ATTI??

When I took flight I confirmed on the controller I was in P-opti mode and had obstacle avoidance on so I could fly indoors? How does it just go to atti mode on its own?
2017-3-13
Use props
Advanced
You need to log in before you can reply Login | Register now

Credit Rules