Determination of orientation

[jepawi]

Hi everyone
Im new to this forum so i hope i do everything right.
I'm a mechanical enginnering student and currently working on a rocket with some collegues. Our ultimate goal is to build an active-stabelized rocket using fin-control. We're struggeling with the determination of the orientation. We already successfully did this with an accelerometer, gyroscope and magnetometer using a sensor-fusion algorithm (mahony) but then realized that we can't use any algorithm like that as there will act a lot of acceleration apart from earths gravity. Does anyone have an idea on how to do it? Ho do they do it in big rockets? How do youtubers like BPS-Space etc do it? We could just use the gyroscope only but are worried that the measurements are not precise enough.
Thanks in advance for any advice.

 

[Alan15578]

Hi everyone
Im new to this forum so i hope i do everything right.
I'm a mechanical enginnering student and currently working on a rocket with some collegues. Our ultimate goal is to build an active-stabelized rocket using fin-control. We're struggeling with the determination of the orientation. We already successfully did this with an accelerometer, gyroscope and magnetometer using a sensor-fusion algorithm (mahony) but then realized that we can't use any algorithm like that as there will act a lot of acceleration apart from earths gravity. Does anyone have an idea on how to do it? Ho do they do it in big rockets? How do youtubers like BPS-Space etc do it? We could just use the gyroscope only but are worried that the measurements are not precise enough.
Thanks in advance for any advice.

star trackers? Tell us about the success that you have already achieved.
This just in: BPS-Space is a youtuber???

 

[jderimig]

Hi everyone
Im new to this forum so i hope i do everything right.
I'm a mechanical enginnering student and currently working on a rocket with some collegues. Our ultimate goal is to build an active-stabelized rocket using fin-control. We're struggeling with the determination of the orientation. We already successfully did this with an accelerometer, gyroscope and magnetometer using a sensor-fusion algorithm (mahony) but then realized that we can't use any algorithm like that as there will act a lot of acceleration apart from earths gravity. Does anyone have an idea on how to do it? Ho do they do it in big rockets? How do youtubers like BPS-Space etc do it? We could just use the gyroscope only but are worried that the measurements are not precise enough.
Thanks in advance for any advice.

Yes, use the Mahony algorithm but set the set the accelerometer values to 0 to the Mahoney filter after detection of launch.

 

[jepawi]

star trackers? Tell us about the success that you have already achieved.
This just in: BPS-Space is a youtuber???

We haven't accomplished that much yet...
We're still in the development phase and currently designing our first prototype. We don't plan to go to space, at frist we just want to get the active stabilisation to work thus star-trackers won't help that much.

 

[jepawi]

Yes, use the Mahony algorithm but set the set the accelerometer values to 0 to the Mahoney filter after detection of launch.

That sounds good. Could we use magnetometer data and fuse it with the data of the gyroscope?

 

[jderimig]

That sounds good. Could we use magnetometer data and fuse it with the data of the gyroscope?

Not with Mahony I think, its been a while since I read the paper. I think he uses the mag to correct the accelerometer. By forcing a=0 you are essentially just integrating the gyro. For a rocket this will work because the drift rate of modern gyro's are low relative to the duration of a rocket boost.

 

[Handeman]

You really didn't communicate what your actual requirements are for the system. I'm assuming you want to use the stabilization to keep the rocket vertical until it's speed is so slow the fins no long work. There is lots of wiggle room is that simple of requirement. How much seeking and oscillation is acceptable. How many degrees off vertical is acceptable. What is an acceptable flight path. You really need to define your requirements before you start designing any prototype. If you don't define the requirements first, you really don't know what you are trying to build. If you don't stick to the requirements, you will run into the bug-a-boo that has killed many project schedules and more budgets, requirements creep.

You have a semester or two to get this working. You may find the requirements too strict, or too loose. You can change them later, but that should be a formal process to change any requirement and an analysis of the change on schedule and budget should go into the decision. Having an upperclassman systems engineering student on your team to track and control the requirements, schedule, budget, etc. not just for the stabilization system, but for the whole rocket, including air frame, propulsion, tracking, data acquisition, recovery, logistics, launch site coordination, etc. is probably a very good idea

As for the stabilization system, I think you need to figure out how accurate your orientation data needs to be to have the physical effects on the rocket to meet the flight path and other requirements you have. You may not need the triple filtered data to accomplish what you need to. There's no return on investment spending time and money on developing the extra filtering if you don't need it. Just integrating the gyro may be more then enough since any errors in the gyro data will tend to increase with time, the effectiveness of your steering fins will also drop off as the rocket slows after burn out. Integrating the actual air speed of the rocket into the control signal amplitudes might be the hardest part, but also key to making it work. Lots of ways to do that, some more effective than others.

The next question is, do you want to compensate for side winds? Do you want to keep the rocket directly over the pad, or is it moving sideways with the wind acceptable as long as it keeps it's vertical flight? I would not worry about side winds until I got a good working stabilization HW/SW solution. At that point you can try to integrate the "stay above the pad" function.

Good luck.

 

[Adrian A]

The way to do it is to use quaternions and propagate the estimated altitude using just the gyros. Modern MEMS gyros are easily accurate enough for this, and example C code for updating the quaternion with gyro data is available with an internet search. it’s a good skill to develop!

 

[Spacedog49Krell]

That sounds good. Could we use magnetometer data and fuse it with the data of the gyroscope?

Try the Madgwick algorithm which uses magnetometer data for orientation. The main issue I have with magnetometer data is the slower output data rate for rocket orientation. I capture data in the 500-1000 Hz range. The integrated 9 DoF chips output magnetometer data in the 20 Hz high resolution to 80 Hz low resolution range. The internal integrated fusion algorithms output in the 20-50 Hz range. Stand-alone magnetometers are two times the integrated output speed. More work to use, but better resolution for active control.

 

[Adrian A]

The other issue with magnetometer data is that it’s difficult to calibrate well enough to actually improve on the accuracy of the gyro-based attitude estimate. It would only really be helpful for a case where the angular rates have exceeded the gyro capability.