1. Jan 9, 2009

### ArixII

Hi,

intro.
Someone said that it in theory there is no difference between theory and practice, but in practice there is! :)

I am a novice with big dreams for learning Control and DSP well. The biggest dream of today is to have my own Autopilot. As I am at the very beginning, I decided to publish things and how I progress toward it in this forum, so that to get your help and ideas and also provide a tutorial for other people interested in the subject. I broke the Autopilot project to some smaller introductory projects, more reachable (at least it seems so):

1. Study positioning: apply a Kalman filter (KF) to encode ONE code wheel.
2. Study localization: Make a TWO Wheeled Inverted Pendulum Robot.
3. Study the vehicle: Make an Autopilot.

Beside, each project is going to have both simulation and real phases. I think that simulation is something between theory and practice!
=================================

1 - learn to use the KF!
To study positioning, I assemble a code wheel (link to material is provided) on a servo motor controlled by a microcontroller, and use a reflector (sensor) to encode it, and pass the signal through a KF to come to accurate positioning information. Clearly, in simulation phase, I must come to a graph that has both the generated through state of the system, the one with white noise added to, and the result of passing it through the KF, to see and feel the effect of the filter.

Here you are with the physical parts (I receive them the next week):
1. Motor: Continuous Rotation Servo
2. Sensor: TCRT5000L
3. Controller: Arduino Duemilanove

Link to all of them, plus the information about code wheels are provided here:
http://washingtonhw.110mb.com/Kalmanexpr.html

I am editing the mathematics for the KF to be used, to publish it here in the next post (tomorrow).

Obviously, what comes next must be a model in Simulink with the 3 parts above plus the KF block. Now I need your helps and ideas:
1) How do you imagine the block-diagram of the system architecture (Control theory).
2) What is better to take as the system state to form the state-space equations: the angular position of the code wheel, or the distance the wheel has traveled (by counting the wheel rotation)?

I will really appreciate your contributions, because actually this first imagination of the right block diagram (sys. architecture) is hardest to me, as it is the earliest step!

Best Regards,
Arash.