What causes the Phantom 4 Pro to drift in yaw (heading) during programmed flights, such as Orbit Mode? This appears to be a common thread in a number of blogs and is a problem we’ve also encountered. As you know, I like to take on the hard problems, think them through and develop solutions. In this blog, I’ll offer ways to measure the offset and a method of compensation.
As I stated above, this is a common problem, but no one that I know of has determined the cause. Please comment if you have a better explanation and I’ll update this blog.
Are other aerial drones similarly affected? Please comment, I’d love to hear from you.
We See Yaw Drift in All of Our Programmed Flights
The yaw drift that we’ve encountered with our Phantom 4 Pro V2 is much more significant than crabbing (please see our April 28, 2019 blog on crabbing). Our data files indicated that the crabbing effect is around ±1.5 degrees, and is largely compensated by the drone’s flight controller. However, our measured yaw offset runs as high as 30 degrees, sometimes more.
Of note, from our data files we plotted the GPS position, which showed the drone stayed on its programmed circular path and its heading was tangent to the circle.
Measuring Yaw Offset
We program almost all of our aerial drone photography sessions, so when the drone’s camera offsets then it’s pretty obvious in the recorded video. A simple method to measure yaw drift is to record a Point of Interest video. That is, to run a circle around a point with the camera pointed at the center. A large radius allows the drone to be operated at maximum speed (we used 1000 feet radius and 21 mph in our test runs), where the drift was quite noticeable.
For example, print out a Google Map of the test site; then graph the video’s centerline of sight at 15-second intervals. You can measure the yaw offset with a ruler for distance, and a protractor for angle. E.g. measure the distance/angle from the centerline to the center point.
Graphical Data Results
Our data set included ten video runs, taken on different days so we had variations in drone speed, wind speed, and wind direction. In almost every case, the yaw drift was affected by both the drone speed and wind speed. One key measurement was the combined speed of the drone, where we found correlation between the maximum yaw drift and the combined air speed of the drone (that is, heading into the wind).
Our graphical analysis suggests that yaw drift can be minimized when both the wind speed and drone speed are less than 10 mph.
- Drone: a 2-year old DJI Phantom 4 Pro V2 with an iPad 9.7-inch tablet. Yaw effects were similar for both DJI Go 4 and Litchi apps.
- Yaw drift appeared to be the same before and after INS and compass calibrations.
- We tried to force the drone’s yaw drift by hovering 5 feet above ground and blowing the drone with a fan. We blew the drone so hard that the camera’s gimbal was pushed into its stops, but it returned to linear after the wind was reduced. The drone’s airframe did tilt into the wind to maintain position, as we would expect, but it didn’t change its yaw (heading).
- We measured yaw drift in circular “Point of Interest” runs, where the drone’s camera was pointed toward the center and the drone airframe was flying sideways into the wind. CCW runs resulted in less yaw drift, so only one run was CW.
In flight, it appears that the drone’s flight controller is adjusting heading as the drone tilts into the headwind. So, if there’s a large headwind, the drone tilts more to maintain its GPS speed and it also yaws to the left. Since the drone’s legs don’t appear in the video, we conclude that the flight controller must be changing the drone’s airframe, not the camera.
Minimizing the Effects of Yaw Offset
- Fly your drone at a speed of less than 10 mph and when the wind is less than 10 mph.
- The yaw offset can be compensated by changing your programmed center point into the wind.
- Fly a larger diameter radius so the desired field of view is around 80% of the frame, then crop down to the desired field of view in post-processing.
Use an Alternative to Orbit Mode
Program your drone with waypoints, enable curves, and place a Point of Interest in the center. With a little practice you can capture a beautiful “orbit” that doesn’t suffer the effects of yaw drift. This has become our preferred method of flying circular patterns.
We have seen this since the last 6 months to a year on several P4P V1’s and a V2. This did not use to happen on the V1’s. Personally I think it is a firmware issue as it all of the sudden just started happening on different drones. This would remove the possibility that it was a compass/GPS/board problem. We primarily fly autonomous mapping missions in which the drone will fly straight to the next programmed waypoint, but the tilt and yaw of the aircraft are off 15 degrees. The camera stays aligned with the drone so the images that should be parallel to something like a roadway show it off parallel. We inserted the images into CAD software and determined that is ALMOST always 15 degrees and is proportional along the flight line depending on the length of that line.
I’ve seen this on an Mavic 2 zoom since the Zoom was released. I’ve mentioned the issue in the forum only to have DJI say “Thank you we’ve referred this to our developers.” Needless to say the fix is to buy an Autel.