Please select Into the mobile phone version | Continue to access the computer ver.
Spark lost signal and flew away
1234
5059 136 2018-6-22
Uploading and Loding Picture ...(0/1)
o(^-^)o
DMX_MT
Second Officer
Malta
Offline

BrilhasMuito Posted at 2018-6-26 07:56
Man,
You know a lot about drones!
Now everything is clear to me. Thanks for your input. I fully agree with you.


I agree with Brilhas !

I saw Mirek the first time on my Compass Error Post. I think he was talking with you Brilhas.

2018-6-26
Use props
JJBspark
Core User of DJI
Flight distance : 12225059 ft
  • >>>
Netherlands
Offline

Northwood Posted at 2018-6-26 10:27
Long read, but good advice.  A few frayed nerves too.  A couple of things come to mind.  And perhaps these have been raised before....

Why dont we have have a switch to enable dropping to Atti manually, so people can learn to fly in that mode in a safe environment?

Hi Northwood,

Height will not change, uses its barometric sensor, thus in ATTI mode height is nothing to worried about. (if ofcourse pitch and roll angles stays within "normal" range)
3 more axis to control yourself.....

cheers
JJB
2018-6-26
Use props
Mirek6
lvl.4
Flight distance : 609724 ft
Canada
Offline

Northwood Posted at 2018-6-26 10:27
Long read, but good advice.  A few frayed nerves too.  A couple of things come to mind.  And perhaps these have been raised before....

Why dont we have have a switch to enable dropping to Atti manually, so people can learn to fly in that mode in a safe environment?

Northwood,

You ask "Why dont we have have a switch to enable dropping to Atti manually, so people can learn to fly in that mode in a safe environment?"

Spark is advertised as a beginner and user friendly drone so DJI decided not to allow ATTI as a choice. I guess correct decision since, in my view, it would cause more problems than prevent problems (with inexperienced users to which it is directed). Unfortunately, this leaves us unprepared when Spark decides to drop to ATTI on its own :-(.

JJB already answered your other question.
2018-6-26
Use props
StuckinNoDak
lvl.2
Flight distance : 151089 ft
United States
Offline

Mirek, show the guy some respect and don’t be so quick to judge. Some of you users suck up to DJI so so much it’s sick. Always claim it’s the pilots fault. I had a flyaway and it was a similar scenerio as this. Wasn’t my error. The Spark is faulty. Glad I have my Mavic now.

Anyone want to buy a new refurbished Spark bundle??!!
2018-6-26
Use props
DMX_MT
Second Officer
Malta
Offline

StuckinNoDak Posted at 2018-6-26 14:36
Mirek, show the guy some respect and don’t be so quick to judge. Some of you users suck up to DJI so so much it’s sick. Always claim it’s the pilots fault. I had a flyaway and it was a similar scenerio as this. Wasn’t my error. The Spark is faulty. Glad I have my Mavic now.

Anyone want to buy a new refurbished Spark bundle??!!


StuckinNoDak,

I am interested in buying your bundle. How much please ?





2018-6-27
Use props
Northwood
lvl.4
Flight distance : 2148596 ft
Canada
Offline

@Mirek @JJB, thanks for the clarification.  Too bad about that switch... perhaps in some non standard firmware some day, after my Care/warranty is dust in the wind



And I saw flashes of this thought in some of the other topics,  if dropped into ATTI mode,  if you then switch over to sport mode, the manual flight controls give you a better fighting chance to recovery control and flight back to yourself, correct?

Thanks again
2018-6-27
Use props
Mirek6
lvl.4
Flight distance : 609724 ft
Canada
Offline

StuckinNoDak Posted at 2018-6-26 14:36
Mirek, show the guy some respect and don’t be so quick to judge. Some of you users suck up to DJI so so much it’s sick. Always claim it’s the pilots fault. I had a flyaway and it was a similar scenerio as this. Wasn’t my error. The Spark is faulty. Glad I have my Mavic now.

Anyone want to buy a new refurbished Spark bundle??!!

StuckinNoDak,

You say: "Mirek, show the guy some respect and don’t be so quick to judge. Some of you users suck up to DJI so so much it’s sick. Always claim it’s the pilots fault."

Did you read all posts with understanding and saw what I described and what happened?
Or did you just scan through them quickly missing most important parts?
Please do go through them again and if you have an urge to reply, please do it more thoughtfully.
2018-6-27
Use props
JimFandango
lvl.4
Flight distance : 130768 ft
Offline

Mirek6 Posted at 2018-6-27 09:01
StuckinNoDak,

You say: "Mirek, show the guy some respect and don’t be so quick to judge. Some of you users suck up to DJI so so much it’s sick. Always claim it’s the pilots fault."

Careful...don't feed the trolls, Mirak.
2018-6-27
Use props
StuckinNoDak
lvl.2
Flight distance : 151089 ft
United States
Offline

Mirek6 Posted at 2018-6-27 09:01
StuckinNoDak,

You say: "Mirek, show the guy some respect and don’t be so quick to judge. Some of you users suck up to DJI so so much it’s sick. Always claim it’s the pilots fault."

I read all the posts and always think carefully before posting.  
2018-6-27
Use props
Mirek6
lvl.4
Flight distance : 609724 ft
Canada
Offline

So we all learn.

Knelson just contacted me via personal message and happily confirmed that DJI indeed accepted his claim as warranty claim and will replace his drone.

This is in line with what I presented in my analysis in this thread (post #83) – I quote:
“Given the above, I strongly believe that irregular gusts of wind blowing diagonally from the back right side of Spark caused elliptical path of return. Since path should have been straight, Spark’s firmware erroneously interpreted it as disagreement between IMU and compass and beginning of the toilet bowl effect thus dropping Spark to ATTI. Within short period of time, Spark was blown out of range of your mobile WiFi and you lost control.”

As a result I did recommend warranty repair case and this is exactly what happened.

Let’s keep an eye on similar cases in the future.  

Cheers!

Mirek
2018-6-28
Use props
JimboHUN
lvl.4
Flight distance : 37156 ft
Offline

Mirek6 Posted at 2018-6-28 11:52
So we all learn.

Knelson just contacted me via personal message and happily confirmed that DJI indeed accepted his claim as warranty claim and will replace his drone.

I just don't get it, why dropping GPS instead of dropping the compass?

GPS should be able to figure out its direction based on the change of coordinates, like in case of car GPS, no?

Cheers,
2018-6-28
Use props
Jakab Gipsz
lvl.4

Hungary
Offline

Northwood Posted at 2018-6-26 10:27
Long read, but good advice.  A few frayed nerves too.  A couple of things come to mind.  And perhaps these have been raised before....

Why dont we have have a switch to enable dropping to Atti manually, so people can learn to fly in that mode in a safe environment?

Hi Nortwood!

There is a solution if you choose to try it:

Only at your own risk if you know what you are doing.
I suggest the following:

- look for open space without close objects
- No wind
- Start with very fine stick motions
- Stay close by while exercising

Good practice!
2018-6-28
Use props
Mirek6
lvl.4
Flight distance : 609724 ft
Canada
Offline

JimboHUN Posted at 2018-6-28 12:25
I just don't get it, why dropping GPS instead of dropping the compass?

GPS should be able to figure out its direction based on the change of coordinates, like in case of car GPS, no?

JumboHUN,

No, drones cannot do that.
Drone can figure out its GPS position - just like your car can. This is not a problem at all.
But when you fly drone, you send it signals in which direction to fly and how fast to fly. When you push on a stick the vector is created which matches the stick direction and degree of your push. This angle determines so called "commanded" position of your drone. This position is calculated based on current physical position, your compass direction (where the north is) and the vector I am talking about. After a short while drone's controller will compare commanded position with its physical GPS position. If they do not match, it will create an "error vector" and correct the drone's course.

This is a very complex and delicate balancing act between compass, IMU sensors such as gyroscope and accelerometer, and GPS system. If error vectors are becoming large (disagreement between the systems I mentioned), the drone will start to spiral with ever increasing radius without input from your stick. It is called toilet bowl effect. It is dangerous and it is very difficult to control the drone once it starts.

This is why it is common practice to drop drone to ATTI mode once the first signs of toilet bowl effect appear or when there is a disagreement between compass, IMU and GPS. It is much easier to control drone manually in ATTI than fighting toilet bowl effect in GPS.

In this particular case, sideways wind put Spark on trajectory which was consistent with toilet bowl effect without actually any mismatch between compass, GPS and IMU. However, because GPS reading of physical position of the drone where not consistent with commanded position, Spark's firmware interpreted it as toilet bowl effect due to sensors mismatch. This was wrong interpretation. Spark, in this case, should not have dropped to ATTI but should have continued on its elliptical course pushed by the wind and constantly adjusting its trajectory.

This is why DJI accepted warranty claim.
2018-6-28
Use props
JimboHUN
lvl.4
Flight distance : 37156 ft
Offline

Thanks, so the issue with my logic is not the position of the drone but the command inputs.

Makes sense but I will need to give this another read with a fresh pair of eyes
Cheers.
2018-6-28
Use props
Mirek6
lvl.4
Flight distance : 609724 ft
Canada
Offline

JimboHUN Posted at 2018-6-28 13:07
Thanks, so the issue with my logic is not the position of the drone but the command inputs.

Makes sense but I will need to give this another read with a fresh pair of eyes

JumbuHun,

You say: "Thanks, so the issue with my logic is not the position of the drone but the command inputs."

Absolutely right. In this case again, when Spark switched to ATTI, its GPS system was working just fine. It was tracking while Spark was being blown away by the wind. You can see it in logs and you can see it by red line on Phantom site map.

If you want to understand a bit better, I invite you to watch this video:


2018-6-28
Use props
marlowe
lvl.4
Flight distance : 300932 ft
United States
Offline

>> But when you fly drone, you send it signals in which direction to fly and how fast to fly. When you push on a stick the vector is created which matches the stick direction and degree of your push. This angle determines so called "commanded" position of your drone. This position is calculated based on current physical position, your compass direction (where the north is) and the vector I am talking about....

Mirek, I know you are an engineer and will understand this. In theory, direction of movement can be determined from the diff of recent GPS positions. An autonomous AC should be able to navigate with GPS data alone, right?
2018-6-28
Use props
Mirek6
lvl.4
Flight distance : 609724 ft
Canada
Offline

marlowe Posted at 2018-6-28 19:00
>> But when you fly drone, you send it signals in which direction to fly and how fast to fly. When you push on a stick the vector is created which matches the stick direction and degree of your push. This angle determines so called "commanded" position of your drone. This position is calculated based on current physical position, your compass direction (where the north is) and the vector I am talking about....

Mirek, I know you are an engineer and will understand this. In theory, direction of movement can be determined from the diff of recent GPS positions. An autonomous AC should be able to navigate with GPS data alone, right?

marlowe,

You write: "Mirek, I know you are an engineer and will understand this. In theory, direction of movement can be determined from the diff of recent GPS positions. An autonomous AC should be able to navigate with GPS data alone, right?"

Well...
So let's say you do not have compass and want to do it with GPS position only. Your Spark acquires GPS. It knows exactly where it is. You move your stick full forward. How does Spark calculate the vector and which GPS position it should be in next second if it does not know which direction it is facing? If it can't calculate its destination (commanded position), how does it know after it arrived at the next point if this point is where it is supposed to be?

However, in the case in this thread (RTH) we do not have this problem. We know the current position and we know destination (home point). So Spark could theoretically fly there by comparing GPS positions along its route. This would not be efficient - it would "hunt" through trial and error.

But... what if acquired GPS is incorrect due to an error in GPS module? If you rely on GPS only you have no safety net. You have no way of knowing if your GPS is fine or not. If you have compass, gyroscope, accelerometer and GPS working together, Spark's controller can and does check consistency between what they are showing. If there is mismatch and controller knows exactly which sensor failed and why, it may attempt to compensate. However, in many situations it is much simpler and perhaps safer for controller not to assume stuff and simply drop to ATTI allowing pilot to take full control.

Also, Spark is a cheap drone. Higher end drones do have a lot of redundancy built in and a lot more checks and balances. In the case we are discussing even Spark should be able to fly home but firmware got fulled by conflicting signals. Hence justified warranty claim DJI rightfully accepted.
2018-6-28
Use props
1234
Advanced
You need to log in before you can reply Login | Register now

Credit Rules