Position Control/Estimation without GPS??

[weg22]

Hi all,

Just wondering if anyone knows of alternatives to getting position information without GPS? I have an indoor helicopter which I have autonomously hovering, but it starts to drift because of drafts or the IMU is not perfectly positioned (i.e. it's not exactly parallel to the ground). Double integrating accelerometer data is prone to error accumulation so that's not an option.

The only thing I can think of doing (excluding indoor GPS) is to point a camera down towards the ground and use some sort of visual odometry algorithm. If anyone else has any other ideas, I'd appreciate it.

Thanks in advance!

[Unterhausen]

If it's the only thing in the room that's moving, the camera idea will work. Otherwise it's not easy. We have 7 robots we are currently struggling with this problem.

Are you using a Kalman filter to estimate the bias and scale errors? I've often wondered about putting a second set of accelerometers and gyros in an IMU since you can get 3 axis accelerometers and 2 axis gyros. So for dual sensors you would need 5 sensor chips total. One thing you might try is using a low pass filter before you stuff the gyro/accelerometer data into the KF. That would mean you would have to sample at higher rates.

[weg22]

I was about to start on the Kalman filter journey, but after probing a few DSP newsgroups...they told me not to bother. It's a waste of time relying on accelerometers for accurate position estimation...even with a Kalman Filter.

"Forget about it. First, the commercial MEMS accelerometers have the accuracy about 0.1m/c^2. Even if you can improve the accuracy by an order of the magnitude by the use of Kalman filter, etc. it is still way below whatever is required for the inertial navigation. Second, the INS is much more complex math then integrating the accelerations. The reason for that is that the Earth is not the inertial system. Your platform is affected by gravity and tide force as well as the centrifugal and Coriolis forces. At the third, a state of the art INS has a runaway about 1km/h. Quite obviously this will not work indoors."

[ampman]

What about Helicommand by Graupner. It uses optical flow, like an optical mouse. It looks down at the ground and detects movement to give an absolute position. It's not to expensive at 300 Euro.