Auto Pilot Module
End of last year marks XA’s third generation of flight controller system. I remember sometime about one year ago a forum that I've been mostly active in was glowing with people sharing their experience and enjoying themselves flying XA lines of quad. There were fairly a lot of newcomers coming in every week and there was more experienced folks contributing to this thread.
So what went wrong? I think there was reasonable expectation shared by not just a few people for XA that it would deliver solid position hold performance. So a lot of guys I know in this thread bought the GPS system. XA’s offering of GPS-S module did not meet this expectation. The problem was multiplied by FC-S and AHRS-S inherent flaw of being susceptible to EMI. Initial performance of GPS-S was so disappointing that a lot of people crashed their copter. People got mad, feeling that they have wasted their money and time.
The GPS-S and compass module was plagued with tons of problem. GPS sat lock was really slow, initial lock in cold boot can take more than 10 minutes. Position hold’s algorithm was terrible, altitude hold doesn’t seem like functioning and Return to Home is more like meteor hitting the earth.
XA seemed to maintain silence and too late to realize the gravity of the situation. After several firmware updates, position hold and RTH were marginally achievable, meaning the quad would hold position in not too windy condition. The setup to make GPS-S to work is not really user friendly as instructions cannot be found in printed material and even when it does work the result is just marginal. When setup properly, it works in calm or slightly windy conditions. And I have posted two videos to proof that, too.
So I was curious to know how autopilot module would perform especially in windy conditions. I’ve posted 4 videos of X450-based quad equipped with autopilot and compass modules doing position hold in windy to very windy conditions.
My initial reaction was that position hold worked. I let go of the stick and in about 1-2 seconds the quad maintains its position. The transition from movement to position hold hovering is not as seamless as DJI WKM as the quad would be slightly drifting for 1-2 seconds before position and altitude hold kicks in. Position hold in horizontal axis was actually pretty good, and because of the vast improvement it had demonstrated over the performance of GPS-S I was quite impressed.
Height hold was there but the windier the flight condition was, the less accurate height hold would perform. Height hold performance is greatly reduced whenever there’s wind gust. It’s accuracy is probably ranging from perfectly level in mid-air to 3-5 meters on vertical axis.
I was just happy that the quad showed efforts and consistency to maintain its position and altitude. With GPS v1 the quad would just drift in very windy conditions.
One thing that I kept in mind when conducting the test was that my test bed was only a 450 class quad with cheap XA motors. My droidworx WKM would fly much much better in similar situations. However, the dx frame is larger than the 450 quad, motors are of much better quality, props are better and the FC is touted as professional grade.
I flew WKM with XA 450 frame and the performance was indeed better comparing to the auto pilot’s. Height hold was exceptional and accuracy on horizontal axis is better, but not much better.
My comment so far is that the autopilot module seems to show promise for those who want to get started to light aerial photography or noobs learning to fly nose in. At this stage of development, it may not be well suited for such aerial video application where still shots are taken from high altitude in very windy conditions for an extended period of time. The quad would be drifting up and down in that kind of scenario. FC-P and AHRS-v2 + Auto Pilot do not and I believe it will never give you pro-level performance but position hold performance is definitely better than several other offerings as shown by videos that I’ve been comparing with. Just check out other threads on FCs.
Height hold performance seem to be a matter of algorithm that needs improving. Overall performance should increase by using better sensors in AHRS-P and algorithmic improvements, but the professional grade AHRS is still not ready to be released by XA.
Here’s what I gather about the autopilot module:
- GPS lock from cold boot is under a minute, about 40ish seconds. With warm boot, gps lock is just a few seconds. Unbelievable. My WKM takes more time to lock sats in warm boot.
- Transitioning from movement to position hold is not seamless, there’s about 1-2 seconds of delay before position and altitude hold kicks in.
- position hold in horizontal axis is actually pretty good, not stellar but it works.
- Height hold works but it appears that change of wind conditions would make the quad bouncy, drifting up or down. When strong gust hits the quad, it would be drifting upward, which is not really good when flying in high altitude and very windy conditions. Would be great if the accuracy of height hold can be increased and height hold is implemented either by flipping a switch or set at 50% of throttle.
- Return to home works but the velocity is still a bit fast, even too fast to my liking. I’d like to see it slowed down just a little bit. No auto landing, guys.
- Carefree and height hold works. Yaw twitch? See my earlier post.
- Compass accuracy for fast forward flight still needs improving, mine is off several degrees to the left in FF. Maybe fixable by repositioning the compass to compensate the deviation, haven’t tried this yet.
- It appears to me that I need to let the quad to sit still for a while after plugging lipo (no more than a minute) otherwise the quad would be drifting forward when doing position hold.
A week ago an XA crew told me that XA is working to improve height hold and overall performance. There’s no ETA for the new autopilot firmware but I am still impressed that XA has greatly improved over the marginal performance of GPS-S. The autopilot module in short, is what GPS-S supposed to be.
http://www.youtube.com/watch?v=Jb2SfL-hx34&context=C32c74e2ADOEgsToPDskJULnI7wbLKD0b2o5ElgtkD
http://www.youtube.com/watch?v=CU88KrvwjUc&context=C32c74e2ADOEgsToPDskJULnI7wbLKD0b2o5ElgtkD
http://www.youtube.com/watch?v=Rxcfj8t6w8k&context=C32c74e2ADOEgsToPDskJULnI7wbLKD0b2o5ElgtkD
http://www.youtube.com/watch?v=ohg1exkSkfQ&context=C32c74e2ADOEgsToPDskJULnI7wbLKD0b2o5ElgtkD
End of last year marks XA’s third generation of flight controller system. I remember sometime about one year ago a forum that I've been mostly active in was glowing with people sharing their experience and enjoying themselves flying XA lines of quad. There were fairly a lot of newcomers coming in every week and there was more experienced folks contributing to this thread.
So what went wrong? I think there was reasonable expectation shared by not just a few people for XA that it would deliver solid position hold performance. So a lot of guys I know in this thread bought the GPS system. XA’s offering of GPS-S module did not meet this expectation. The problem was multiplied by FC-S and AHRS-S inherent flaw of being susceptible to EMI. Initial performance of GPS-S was so disappointing that a lot of people crashed their copter. People got mad, feeling that they have wasted their money and time.
The GPS-S and compass module was plagued with tons of problem. GPS sat lock was really slow, initial lock in cold boot can take more than 10 minutes. Position hold’s algorithm was terrible, altitude hold doesn’t seem like functioning and Return to Home is more like meteor hitting the earth.
XA seemed to maintain silence and too late to realize the gravity of the situation. After several firmware updates, position hold and RTH were marginally achievable, meaning the quad would hold position in not too windy condition. The setup to make GPS-S to work is not really user friendly as instructions cannot be found in printed material and even when it does work the result is just marginal. When setup properly, it works in calm or slightly windy conditions. And I have posted two videos to proof that, too.
So I was curious to know how autopilot module would perform especially in windy conditions. I’ve posted 4 videos of X450-based quad equipped with autopilot and compass modules doing position hold in windy to very windy conditions.
My initial reaction was that position hold worked. I let go of the stick and in about 1-2 seconds the quad maintains its position. The transition from movement to position hold hovering is not as seamless as DJI WKM as the quad would be slightly drifting for 1-2 seconds before position and altitude hold kicks in. Position hold in horizontal axis was actually pretty good, and because of the vast improvement it had demonstrated over the performance of GPS-S I was quite impressed.
Height hold was there but the windier the flight condition was, the less accurate height hold would perform. Height hold performance is greatly reduced whenever there’s wind gust. It’s accuracy is probably ranging from perfectly level in mid-air to 3-5 meters on vertical axis.
I was just happy that the quad showed efforts and consistency to maintain its position and altitude. With GPS v1 the quad would just drift in very windy conditions.
One thing that I kept in mind when conducting the test was that my test bed was only a 450 class quad with cheap XA motors. My droidworx WKM would fly much much better in similar situations. However, the dx frame is larger than the 450 quad, motors are of much better quality, props are better and the FC is touted as professional grade.
I flew WKM with XA 450 frame and the performance was indeed better comparing to the auto pilot’s. Height hold was exceptional and accuracy on horizontal axis is better, but not much better.
My comment so far is that the autopilot module seems to show promise for those who want to get started to light aerial photography or noobs learning to fly nose in. At this stage of development, it may not be well suited for such aerial video application where still shots are taken from high altitude in very windy conditions for an extended period of time. The quad would be drifting up and down in that kind of scenario. FC-P and AHRS-v2 + Auto Pilot do not and I believe it will never give you pro-level performance but position hold performance is definitely better than several other offerings as shown by videos that I’ve been comparing with. Just check out other threads on FCs.
Height hold performance seem to be a matter of algorithm that needs improving. Overall performance should increase by using better sensors in AHRS-P and algorithmic improvements, but the professional grade AHRS is still not ready to be released by XA.
Here’s what I gather about the autopilot module:
- GPS lock from cold boot is under a minute, about 40ish seconds. With warm boot, gps lock is just a few seconds. Unbelievable. My WKM takes more time to lock sats in warm boot.
- Transitioning from movement to position hold is not seamless, there’s about 1-2 seconds of delay before position and altitude hold kicks in.
- position hold in horizontal axis is actually pretty good, not stellar but it works.
- Height hold works but it appears that change of wind conditions would make the quad bouncy, drifting up or down. When strong gust hits the quad, it would be drifting upward, which is not really good when flying in high altitude and very windy conditions. Would be great if the accuracy of height hold can be increased and height hold is implemented either by flipping a switch or set at 50% of throttle.
- Return to home works but the velocity is still a bit fast, even too fast to my liking. I’d like to see it slowed down just a little bit. No auto landing, guys.
- Carefree and height hold works. Yaw twitch? See my earlier post.
- Compass accuracy for fast forward flight still needs improving, mine is off several degrees to the left in FF. Maybe fixable by repositioning the compass to compensate the deviation, haven’t tried this yet.
- It appears to me that I need to let the quad to sit still for a while after plugging lipo (no more than a minute) otherwise the quad would be drifting forward when doing position hold.
A week ago an XA crew told me that XA is working to improve height hold and overall performance. There’s no ETA for the new autopilot firmware but I am still impressed that XA has greatly improved over the marginal performance of GPS-S. The autopilot module in short, is what GPS-S supposed to be.
http://www.youtube.com/watch?v=Jb2SfL-hx34&context=C32c74e2ADOEgsToPDskJULnI7wbLKD0b2o5ElgtkD
http://www.youtube.com/watch?v=CU88KrvwjUc&context=C32c74e2ADOEgsToPDskJULnI7wbLKD0b2o5ElgtkD
http://www.youtube.com/watch?v=Rxcfj8t6w8k&context=C32c74e2ADOEgsToPDskJULnI7wbLKD0b2o5ElgtkD
http://www.youtube.com/watch?v=ohg1exkSkfQ&context=C32c74e2ADOEgsToPDskJULnI7wbLKD0b2o5ElgtkD