Skydio 2 Questions

[aerial_photos]

i received my Skydio 2 pro last week and have flown it for about 2 hours now. I am impressed with its ability to ”know” its heading but puzzled as to how it does it. I too sent a help query to Skydio and got the reply shown below. How can there be a magnetometer with such a powerful magnet for the battery attachment? If there is no magnetometer how does if figure out which way it’s pointing?

John (Skydio)

Oct 23, 2020, 6:14 PM PDT

Hi Jeff,

Sorry I misunderstood the question, there is a onboard magnetometer contributing to the Skydio 2's autonomy system. Currently that is not a telemetry reading on the application, but we......

 

[LivinLarge]

i received my Skydio 2 pro last week and have flown it for about 2 hours now. I am impressed with its ability to ”know” its heading but puzzled as to how it does it. I too sent a help query to Skydio and got the reply shown below. How can there be a magnetometer with such a powerful magnet for the battery attachment? If there is no magnetometer how does if figure out which way it’s pointing?

John (Skydio)

Oct 23, 2020, 6:14 PM PDT

Hi Jeff,

Sorry I misunderstood the question, there is a onboard magnetometer contributing to the Skydio 2's autonomy system. Currently that is not a telemetry reading on the application, but we......

Heading is easily calculated from changes in GPS position. You may have noticed that the navigation system in your car does not know your orientation until you move. I expect this works the same way.

 

[VEGASROBBI]

Can you post the remainder of the reply? Sounds like the S2 may have a compass I did not think it did.

Heading can be determined with GPS and optical used for Follow.

If the S2 has a compass maybe it is calibrated to compensate for the battery?

 

[LivinLarge]

Can you post the remainder of the reply? Sounds like the S2 may have a compass I did not think it did.

Heading can be determined with GPS and optical used for Follow.

If the S2 has a compass maybe it is calibrated to compensate for the battery?

Considering that there is no calibration procedure when every other drone requires it, makes me think there is no compass.

 

[skydioconvert]

Considering that there is no calibration procedure when every other drone requires it, makes me think there is no compass.

Have to agree with you about there being no compass.... the fact that the S2 flies unaffected near large iron structures further suggests no compass is on board.

 

[VEGASROBBI]

While I agree I'd like to see the rest of the reply?


Hi Jeff,

Sorry I misunderstood the question, there is a onboard magnetometer contributing to the Skydio 2's autonomy system. Currently that is not a telemetry reading on the application, but we......

 

[aerial_photos]

For GPS to be used for heading the drone needs to be moving. If I yaw it in a fixed spot (no translation) it still shows a plausible heading on the FPV. So how can it do that just using GPS alone?

 

[LivinLarge]

For GPS to be used for heading the drone needs to be moving. If I yaw it in a fixed spot (no translation) it still shows a plausible heading on the FPV. So how can it do that just using GPS alone?

Skydio seems to always fly forward a bit immediately after launch. This may be to set the initial heading. Most drones include gyros and accelerometers which help hold position and orientation as well as detecting yaw. This can accumulate errors over time.
I do wonder if you rotated it around several full circles, if the orientation on the screen would still be accurate.

 

[aerial_photos]

While I agree I'd like to see the rest of the reply?


Hi Jeff,

Sorry I misunderstood the question, there is a onboard magnetometer contributing to the Skydio 2's autonomy system. Currently that is not a telemetry reading on the application, but we......

Currently that is not a telemetry reading on the application, but we do have a more enterprise focused application that will offer more telemetry data coming before the end of this year.

 

[aerial_photos]

Skydio seems to always fly forward a bit immediately after launch. This may be to set the initial heading. Most drones include gyros and accelerometers which help hold position and orientation as well as detecting yaw. This can accumulate errors over time.
I do wonder if you rotated it around several full circles, if the orientation on the screen would still be accurate.

Exactly what I plan to do, if I can’t fool it after several aggressive yaws in place I guess there must be some other magic going on

 

[Ridefreak]

When using the beacon and app, if you're launch the drone it takes off then turns around and gets a visual lock on the subject and quickly soon after it shows GPS indication around the subject. At that point it knows it's GPS coord and also the Beacon's but often it doesn't know north from south. If you change it's orientation using the arrows on the beacon (which require polar orientation to determine where it should be in relation to the subject) you'll see that most of the time it won't change until there's movement in some direction and it established orientation. That's because it was just powered up and hasn't traveled far enough to figure that out but sometimes there's enough movement that it does. If you come to a stop after it's followed the subject and do the same thing it will swing around like you'd expect. Pretty evident it's using GPS to establish it's polar orientation.

 

[VEGASROBBI]

The mystery remains, the rep may have been in error. They mentioned enterprise which may have a compass.

My experience is the same as the others you need to be moving to get a heading lock. That works fine for me because I’m always in follow, the advantages of flying near metal objects is a definite advantage.

 

[aerial_photos]

The Skydio rep admitted his error in claiming the Skydio 2 had an onboard magnetometer. After a few more exchanges I received this message from him.
shown below. Based on a test I did yesterday the Skydio 2 has no clue which way it is pointing on the ground before takeoff but once airborne it does determine its heading after a few seconds even with just a little lateral motion. I guess it uses visual odometry and inertial sensors to “keep track” of its heading after translation stops and it’s just yawing around in the same location.

Hi Jeff,

Well, the best way to explain it is that Skydio 2 uses visual inertial odometry and an internal IMU to understand its position for state planning coupled with GPS and information collected from the Skydio 2 app as well. I guess I didn't quite realize that you wanted an explanation of how Skydio 2 understands its position. If you have any more questions I'll be glad to answer them.
Best regards,
John

 

[aerial_photos]

Exactly what I plan to do, if I can’t fool it after several aggressive yaws in place I guess there must be some other magic going on

It did show a valid heading after I did a bunch of fast yaw in place maneuvers left and right. So I guess the vision and inertial sensors a pretty good at keeping track of its heading in the absence of lateral motion.

 

[LivinLarge]

The Skydio rep admitted his error in claiming the Skydio 2 had an onboard magnetometer. After a few more exchanges I received this message from him.
shown below. Based on a test I did yesterday the Skydio 2 has no clue which way it is pointing on the ground before takeoff but once airborne it does determine its heading after a few seconds even with just a little lateral motion. I guess it uses visual odometry and inertial sensors to “keep track” of its heading after translation stops and it’s just yawing around in the same location.

Hi Jeff,

Well, the best way to explain it is that Skydio 2 uses visual inertial odometry and an internal IMU to understand its position for state planning coupled with GPS and information collected from the Skydio 2 app as well. I guess I didn't quite realize that you wanted an explanation of how Skydio 2 understands its position. If you have any more questions I'll be glad to answer them.
Best regards,
John

That is what I expected. Glad to hear that it does well after multiple yaws.

 

[skydioconvert]

If you trust Google:
Odometry is the use of data from motion sensors to estimate change in position over time. It is used in robotics by some legged or wheeled robots to estimate their position relative to a starting location. Wikipedia

Visual odometry is the process of determining equivalent odometry information using sequential camera images to estimate the distance traveled.



OK.... visual odometry, an internal IMU and GPS all help to determine distance and direction traveled BUT how do these inputs determine initial orientation??? ... None of those inputs provide information to determine the initial direction the drone is pointed like a compass heading would.

 

[Ridefreak]

The Beacon and the SD both have a GPS, they communicate with each other, it takes off an points itself at the beacon's location, easy since it's a Lat/Long and it's aware of it's own Lat/Long. My other drone doesn't have a compass ether but it also has two GPSs and operates similar to the SD. Watching both of them right after takeoff is almost identical. They point themselves at their tether.

 

[LivinLarge]

If you trust Google:
Odometry is the use of data from motion sensors to estimate change in position over time. It is used in robotics by some legged or wheeled robots to estimate their position relative to a starting location. Wikipedia

Visual odometry is the process of determining equivalent odometry information using sequential camera images to estimate the distance traveled.



OK.... visual odometry, an internal IMU and GPS all help to determine distance and direction traveled BUT how do these inputs determine initial orientation??? ... None of those inputs provide information to determine the initial direction the drone is pointed like a compass heading would.

Visual odometry and an IMU can hold a heading to keep the yaw from drifting. Initially the Skydio will not know its heading until it moves. It normally flies a short distance forward on take off. Knowing the starting point and ending point from GPS, it can easily calculate which way it is pointed. Once that heading is known, the IMU and visual odometry can measure changes in heading and update as it yaws.

 

[VEGASROBBI]

The Skydio rep admitted his error in claiming the Skydio 2 had an onboard magnetometer. After a few more exchanges I received this message from him.
shown below. Based on a test I did yesterday the Skydio 2 has no clue which way it is pointing on the ground before takeoff but once airborne it does determine its heading after a few seconds even with just a little lateral motion. I guess it uses visual odometry and inertial sensors to “keep track” of its heading after translation stops and it’s just yawing around in the same location.

Hi Jeff,

Well, the best way to explain it is that Skydio 2 uses visual inertial odometry and an internal IMU to understand its position for state planning coupled with GPS and information collected from the Skydio 2 app as well. I guess I didn't quite realize that you wanted an explanation of how Skydio 2 understands its position. If you have any more questions I'll be glad to answer them.
Best regards,
John

 

[skydioconvert]

The Beacon and the SD both have a GPS, they communicate with each other, it takes off an points itself at the beacon's location, easy since it's a Lat/Long and it's aware of it's own Lat/Long. My other done doesn't have a compass ether but it also has two GPSs and operates similar to the SD. Watching both of them right after takeoff is almost identical. They point themselves at their tether.

Ok, but if you don't have/use a beacon, from where does the 2nd GPS input come from?

EDIT: I think LivinLarge's last post answers that question.