mavic 2 GPS vs Phantom 3S

[BudWalker]

the zip file is the first battery flight. I think the first RTH was dead nuts on. Don't remember how many RTH I did. The DAT file is the last of the day. I had to zip the first (too big) so I don't know if that will mess with any data.

I'm particularly interested in seeing the .DAT were you reported that the navHealth data couldn't be plotted. Both the .DATs you submitted didn't seem to have this problem.

There were 7 or 8 landings in the two .DATs, most of them RTH landings. Except for the first landing in FLY098 they were normal. The RTH landings landed within .5 meter of the HP. There were 2 landings under pilot control that landed outside a .5 meter distance from the HP.

If you want to pursue investigating the problem you stated please provide a .DAT where it happened. If there are multiple landings then it will be necessary to know which landing.

There doesn't seem to be a problem with the GPS. DOP values are low and the navHealth values are either 4 or 5 indicating the FC has confidence in the lat/long values computed using the GPS and IMU data.

An RTH was commanded in the first landing in FLY098. The M2 moved directly over the HP (at time 358 secs) but then moved laterally (mostly south) as it first yawed to the launch orientation and then descended.

The straight red line is the return to the HP (yellow H symbol) and the squiggly red line is the movement to the south after arriving directly over the HP.

I'm guessing the PL system liked the spot where it landed better than the launch site.

I did consider the possibility that the GPS data was incorrect. That the M2 actually landed at the HP but the GPS data was showing that it landed south of the HP. But, from the accelerometer and gyro data it's apparent that you picked up the M2 and moved it back closer to the HP.

 

[paulatkin73]

For a start, your idea about the accuracy of the P3S isn't realistic and it seems your idea about GPS accuracy in general isn't correct.
GPS, any GPS is not pinpoint accurate and its inaccuracy will be variable.
Most of the time (under good conditions) it will give a position fix within 2 metres but sometimes it will be even be outside that.
Your P3S could not consistently perform an RTH landing within 1 metre, as the GPS equipment is not capable of that degree of accuracy.

Your Mavic 2 should be capable of similar GPS accuracy to any P3 and if you really believe it isn't, you need to look at the recorded flight data to see if there really is an issue and what might be causing it.

And finally, the autolanding following RTH is just a gimmick anyway, rather than an important feature.
A human should be able to cancel RTH and land properly every time.

without having access to source code it is all just a guess work.
i do not exclude a possibility that they coded to use GPS to navigate to a home point only, at altitude, and then as it switches into landing mode, they go by the visual system only for precision landing and ignore gps coordinates - that would explain possible deviation from the exact take off point.

in practical reality most likely the reason of the an offset during landing is a lack of attention given by the operator to the HDOP value during a takeoff, so, gps coords mutate after vehicle gets better gps lock.

 

[BudWalker]

......

in practical reality most likely the reason of the an offset during landing is a lack of attention given by the operator to the HDOP value during a takeoff, so, gps coords mutate after vehicle gets better gps lock.

There is bad news and good news. The bad news is that HDOP isn't available and can't be checked by the pilot. The good news is that gpsLevel which depends on the HDOP data is available. Here it's shown as 4 (out of 5)


GpsLevel is a measure of the confidence that the flight controller has in the lat/long values that are being computed from IMU, GPS and magnetometer data. As such, low DOP values are required for a gpsLevel of 4 or 5.

 

[ruggb]

Bud
where can I find tutorials on analyzing telemetry data? It seems to be rather scarce.

 

[BudWalker]

Bud
where can I find tutorials on analyzing telemetry data? It seems to be rather scarce.

You could try
Flight Controller Data Analysis Series Tutorials--Part 1

You might want to try using CsvView instead of the DJI Viewer that's used in that tutorial.

 

[Former Member]

thunderdrones
the top strobe has been relocated to that position from 3" back

I am sure that's a better position for it. Nothing under there but plastic.

 

[Keule]

where can I find tutorials on analyzing telemetry data?

Here's some documentation from DJI on this topic:

 

[Saigondroner]

I had a Phantom 3S. When it did an RTH the landing was almost always within 1m of takeoff point. That is by GPS location only.
My new M2Z only lands on my 1/2m pad maybe 50% of the time. I have very carefully calibrated IMU & compass several times to make sure I got it right. When I takeoff, I hover for a photo op. When it arrives at "home" (@200ft) it is not over the landing pad. Sometimes it is off by 4-5m or more. At that distance it can't photo correct when it get to ground level. If it is off only 1-2m it does photo correct at about 5m and then lands on the pad.

Both were flying in the same location.
If the Phantom GPS alone brought it down on target, what is wrong with the M2Z GPS?

IMHO, DJI has sacrificed basic nav and telymmetry features to sell drones to a larger market. Tapfly vs. RTH? I also believe they have been so successful selling their drones, they have been forced to geofence, etc. to minimize errors.

I'd fly a P3* over most DJI birds theses days. I have an original P2, however....it is a handful

 

[sar104]

IMHO, DJI has sacrificed basic nav and telymmetry features to sell drones to a larger market. Tapfly vs. RTH? I also believe they have been so successful selling their drones, they have been forced to geofence, etc. to minimize errors.

I'd fly a P3* over most DJI birds theses days. I have an original P2, however....it is a handful

What do you think they have sacrificed in terms of navigation and telemetry? Home lock and course lock are the only functions that I can think of that have gone away. Everything else seems to keep improving.

 

[Former Member]

I have an original P2, however....it is a handful

I think everyone should own a P2. It would make them appreciate their Mavics more.

 

[sar104]

I think everyone should own a P2. It would make them appreciate their Mavics more.

I especially enjoyed "Manual" mode. At least for the several seconds of impending disaster before I had to bail back to P-GPS.

 

[Saigondroner]

What do you think they have sacrificed in terms of navigation and telemetry? Home lock and course lock are the only functions that I can think of that have gone away. Everything else seems to keep improving.

I think everyone should own a P2. It would make them appreciate their Mavics more.

I agree about P2 ownership. I have the Gimbal off my P2, with only an OSD installed, so I can see battery status, etc. The Mav2P failure seem to be much more prevalent than other DJI drones before. I have not experienced any serious problems, but the included Smart Controller was a big mistake. I've have given up updating the SC, it has failed dozens of times, 2-3 machines tried. I cannot use Litchi, or other autonomous software (except DJI's).

For certain, newer models are easier to fly, control and land. My Spark & Air have many software problems...sometime hardware pointing to software, and vice versa. I pretty much keep the M2P shelved, and fly a Mavic 1, and never update the firmware. It seems that they are suffering from bloatware more than anything. Geofencing, forced login, has a lot to with my very subjective opinon.

 

[sar104]

I agree about P2 ownership. I have the Gimbal off my P2, with only an OSD installed, so I can see battery status, etc. The Mav2P failure seem to be much more prevalent than other DJI drones before. I have not experienced any serious problems, but the included Smart Controller was a big mistake. I've have given up updating the SC, it has failed dozens of times, 2-3 machines tried. I cannot use Litchi, or other autonomous software (except DJI's).

For certain, newer models are easier to fly, control and land. My Spark & Air have many software problems...sometime hardware pointing to software, and vice versa. I pretty much keep the M2P shelved, and fly a Mavic 1, and never update the firmware. It seems that they are suffering from bloatware more than anything. Geofencing, forced login, has a lot to with my very subjective opinon.

Most of that is incorrect. With the assumption that the rate of reporting of incidents is approximately constant, and independent of model, the M2 is not suffering from higher failure rates. As for the SC, DJI just released the necessary SDK, so Litchi should be able to run on it as soon as an update is released.

I don't know anyone who has flown both the MP and the M2 and prefers the MP. It was a fine aircraft, but it's not in the same league as the M2. So your M2 hasn't had any problems but you don't fly it? Why? Is it just that you can hack the MP on old firmware much more easily?

 

[Saigondroner]

I don’t trust a controller that complains it needs an update, and has been impossible to rectify, to date.

I have never called DJI support. I hear nothing but bad tales. I bought from a supposed authorized dealer, who lowered the cash price 10% when I asked for a red (official) receipt.

I knew what I was doing, know the game in Asia, took my chances. If I had a defective unit, I’d bite the bullet and replace it myself. The SC is really more of an endemic problem.

I don’t care for DJI’s marketing hype, which has devolved into bold faced lies.

My bad for not having done due diligence when buying the M2. No customer should be expected to spend such an amount, and then told to wait 3 months, before an SDK is published. DJI has oversold several products, and is getting worse.

DJI is a ~4billion USD company, and adopted the Apple strategy well.

 

[sar104]

I don’t trust a controller that complains it needs an update, and has been impossible to rectify, to date.

I have never called DJI support. I hear nothing but bad tales. I bought from a supposed authorized dealer, who lowered the cash price 10% when I asked for a red (official) receipt.

I knew what I was doing, know the game in Asia, took my chances. If I had a defective unit, I’d bite the bullet and replace it myself. The SC is really more of an endemic problem.

We are not going to agree on most of this.

I've never called DJI support either, but I can say that in most of the cases that I've helped out with, which is now quite a few, DJI appears to have been fair and helpful. Your conclusion, based on nothing but "tales", doesn't seem reasonable. Most of the tales on these forums complaining about DJI turn out to be from people trying to turn pilot error into DJI's fault, who then get frustrated when DJI doesn't roll over and give them free stuff. What does your shady dealer have to do with the argument?

I don’t care for DJI’s marketing hype, which has devolved into bold faced lies.

Such as?

My bad for not having done due diligence when buying the M2. No customer should be expected to spend such an amount, and then told to wait 3 months, before an SDK is published. DJI has oversold several products, and is getting worse.

That does seem like your bad. So DJI advertised that it would work with Litchi?

DJI is a ~4billion USD company, and adopted the Apple strategy well.

You mean the strategy that has resulted in tens of millions of satisfied, loyal customers? That seems like a fine model to copy.

 

[paulatkin73]

I agree about P2 ownership. I have the Gimbal off my P2, with only an OSD installed, so I can see battery status, etc. The Mav2P failure seem to be much more prevalent than other DJI drones before. I have not experienced any serious problems, but the included Smart Controller was a big mistake. I've have given up updating the SC, it has failed dozens of times, 2-3 machines tried. I cannot use Litchi, or other autonomous software (except DJI's).

For certain, newer models are easier to fly, control and land. My Spark & Air have many software problems...sometime hardware pointing to software, and vice versa. I pretty much keep the M2P shelved, and fly a Mavic 1, and never update the firmware. It seems that they are suffering from bloatware more than anything. Geofencing, forced login, has a lot to with my very subjective opinon.

If you need help to sort out smart controller update - ask in a different topic. I have one and it is all doable.
M2p flies and works fine. No need to escalate needless panic.

 

[Saigondroner]

If you need help to sort out smart controller update - ask in a different topic. I have one and it is all doable.
M2p flies and works fine. No need to escalate needless panic.

Point well taken.Yes, the M2P flies quite well....however..how does Litchi loaded on the SC run for you?

 

[Saigondroner]

I especially enjoyed "Manual" mode. At least for the several seconds of impending disaster before I had to bail back to P-GPS.

Oh yeah, ATTI, IOC off. light the fuse...go

 

[Former Member]

Point well taken.Yes, the M2P flies quite well....however..how does Litchi loaded on the SC run for you?

I think DJI just released the sdk for the SC, so no LItchi yet for the SC.

 

[Saigondroner]

Most of that is incorrect. With the assumption that the rate of reporting of incidents is approximately constant, and independent of model, the M2 is not suffering from higher failure rates. As for the SC, DJI just released the necessary SDK, so Litchi should be able to run on it as soon as an update is released.

I don't know anyone who has flown both the MP and the M2 and prefers the MP. It was a fine aircraft, but it's not in the same league as the M2. So your M2 hasn't had any problems but you don't fly it? Why? Is it just that you can hack the MP on old firmware much more easily?

I understand and respect your opinion, My M2 cannot fly autonomously unless I redo dozens of missions from Litchi to DJI's substandard version. I have no interest in 'hacking firmware'. I simply want a product to perform as advertised. The controller is a decent percentage of the product. I hear the DJI is finally releasing API's and the SDK. I eagerly await that event.