I am looking at the Sparkfun version of the MPU9150, which should be a good start for building an inertial nav unit. The MPU9150 datasheet describes the built in Digital Motion Processor, but I don't see more than a vague explanation of what that does. I would hope that it puts out X, Y, Z coordinates corrected for tilt, but I don't find any indication that that is true.
If you have some hands-on experience with this, and can explain the Digital Motion Processor, I'd certainly appreciate any information.
Thanks in advance.
John Doner
I'm interested too, but I'm looking at the ones on ebay for $12. Currently I've got one of these on order 10DOF IMU MPU6050 HMC5883L BMP085 Gyroscope Acceleration Compass Module Arduino | eBay
It's got an MPU6050 HMC5883L BMP085 IMU on order, I have not used the MPU6050 I hope its alight. I'm currently testing out the L3G4200D and BMA180, but the L3G4200D has issues with spikes I've not figured out.
I'm tempted on ordering one of those with the L3G4200D, but I want it to have a pressure sensor to go along.
The so-called DMP on the MPU6050 and MPU9150 will do so-called six-dof motion fusion, which is not very useful. The manufacturer's advertising tends to overstate its capabilities. Furthermore, it has to be programmed using secret microcode which they have not published.
Strange: I wonder how much to have to pay to learn the secret handshake?
But another question: I have not done motion tracking before, although I do have a Ph.D. in mathematics, so so project does not scare me. Why do you say 6 DOF motion tracking "is not very useful"? It intuitively seems to me that if you account for all accelerometer readings, with all rotations of the accelerometers taken into account, it should be possible to at least theoretically keep perfect track of 3D motion.
Again, thanks for any info.
John Doner
It intuitively seems to me that if you account for all accelerometer readings, with all rotations of the accelerometers taken into account, it should be possible to at least theoretically keep perfect track of 3D motion.
True, but if you take into account the inaccuracy of consumer-grade devices, you will discover that they are nearly useless for inertial navigation.
A major problem is the inability to determine the exact orientation of the gravity vector. Here is a nice summary of the consequences http://www.chrobotics.com/library/accel-position-velocity
(Beware of numerical errors in the acceleration and position, but not velocity, columns of Table 1 -- I contacted Caleb at CHRobotics and he promised to correct them.)
Edit: I now see that this is at least the second time you have posted this question, but you don't seem to be interested enough in the responses to read up on the topic.
I'm not sure why you think I'm not interested enough to read up on the topics I ask questions about. I am always grateful for the responses I get, and I am very happy that this forum exists.
I lead a busy life with numerous research threads going on simultaneously. I appreciate your help, but I may not promptly respond to suggestions, and I am not sure how you know whether I have responded or not. perhaps you work for the NSA?
John Doner
I'm not sure why you think I'm not interested enough to read up on the topics I ask questions about.
It is obvious that you did not bother to consult the link that I posted, the first time I responded to your question. You are wasting everyone's time. But, I now understand that you are very busy!
You fail the reading test yourself. I did not ask about computational aspects of inertial nav a second time. The first question was on that subject, the second question concerned the MPU9150 chip's capabilities. In my opinion I'm not wasting your time, you're wasting your own time.
Cheers
John Doner