Frank,

First, I really enjoyed your four messages regarding rocket guidance. Do you mind if I make them public? They are fascinating reading.

The short answer to your question about using the UAV DevBoard (UDB) for rocket guidance, is probably not. There certainly are some challenges. I need to think about it, I may get some ideas. For sure, we would have to make some major changes to the firmware, might even have to scrap the hardware design and start over. Here are some things to think about.

1. The firmware for the board right now depends on gravity to correct for gyro drift. That works great for an airplane. For a rocket, I would abandon the accelerometers all together and use the GPS velocity vector to correct for gyro drift. I think it might work just great, especially at the high speeds of a rocket.

2. The LISY gyros are vulnerable to vibration, and acceleration as well. We would want to abandon them and find something else.

Best regards,
Bill

Frank,

First, I thought I might explain a little bit more why the combination of accelerometers and gyros works well for an airplane, somewhat for a helicopter, and not much for a rocket.

The reason is, what an accelerometer really measures is the aerodynamic forces. Since we need them only for drift correction of the gyros, an instantaneous error is ok, but on the average, we need the right answer.

For an airplane, the average of the aerodynamice forces supports the plane, so it is a good indicator, on averages, of where is down.

However, I would think that the aerodynamic forces on a rocket are small during much of its trajectory. Therefore, the accelerometers will not help.

What is needed is something with good average accuracy. Assuming (you tell me if I am wrong) that through the later phases in a rocket's trajectory, its motion through the air causes it to move in the direction that it is pointed. If so, the GPS could be used until it stops working. The EM406 is good up to just beyond 500 meters per second, so it could do the job, if my assumption is correct.

The key would be to get the rocket level prior to launch. The accelerometers would then drift-correct the gyros prior to launch.

At launch, you would turn off the accelerometers for drift correction, switch to GPS. During the early portion of the flight, you would not really need any drift correction anyway. It would kick in from the GPS just about the time you would need it.

I am not sure if the magnetometers would help or not, because in the general case you need two vectors to correct the gyros, and a magnetometer gives you only one. The reason we can get away with only one vector if we use GPS, is because of the assumption that the rocket moves in the direction it is pointed, so we do not need the yaw.

However, in the case of the a magnetometer, the magnetic field has a vertical and a horizontal component, so unless you had a second vector, the fact that you do not know the yaw angle of the rocket would make it impossible to drift-compensate with just a magnetometer.

If you use GPS, gyro, and magnetometer, you could get all 9 components of the direction cosine matrix. That would probably be your best bet. You could do it, I think.

I am not sure what the "big boys" do. I think they simply use very good, low drift gyros. I think that is what they did on the Apollo missions.

By the way, if you have not read it, I highly recommend the book, "Digital Appolo, Human and Machine in Spaceflight", by David A. Mindell, its a really good read, and touches on some of the issues you are interested in.

I also recommend "Journey to the Moon: The History of the Apollo Guidance Computer", by Eldon C. Hall.

Best regards,
Bill

Frank,

It sounds to me like they are using low drift gyros, so you can get away without doing drift compensation. That greatly simplifies the problem, so you do not need GPS or magnetometers.

What it sounds like that they are doing is to use the gyros to get orientation. Then, you assume gravity, and use the accelerometers to compute acceleration.

Another factor that I may be overlooking is the duration of the flight. The shorter the flight, the less of an issue gyro drift is going to be.

Tell me again, how long do you need to have an accurate answer, and what accuracy do you need?

If you had an accurate enough set of gyros and accelerometers, you could use the DCM algorithm to compute everything.

Best regards,
Bill Premerlani

Frank,

One minute is not very long, so you can ignore compensating for gyro drift all together. You do not need a GPS and you do not need a magnetometer.

What you do need are gyros and accelerometers that are accurate enough for what you want to do, and which will operate up to 18 gs.

The gyros and accelerometers on the UAV DevBoard will not do the job for you, but you could design your own board, most likely you could use the IMU that you mentioned. You certainly could do what you outlined in your other comments.

Let me know if you decide to go forward, I could give you a hand.

Best regards,
Bill

Hi Frank,

I am glad you enjoyed the podcast.

Regarding the gyros, I think there is a very good chance the Invensense gyros (IDG-500) will do what you want. In any case, if I were you, I would certainly invest $150 to find out.

Best regards,
Bill