Have gotten most of the algorithm done and are now working on getting it implemented on our product. The first batch of prototypes are due shortly. Unfortunately there are some problems with the current firmware on our box. More specifically the way we can use the I2C bus to communicate to the acc/gyro chip is waaaaaaay to slow at the moment. roughly 3 byte per second. And one reading consists of 2 byte for each of the axis (3 for the accelerometer and 3 for the gyro) plus 1 for temperature. so that's 13 byte per reading; we need minimum 30 readings pr. second! The firmware is created in China and this creates a quite significant language and culture barrier. So it's going to be interesting to see when (if ever) we can get this bug fixed...

Been reading up A LOT on vector and matrix maths to be able to mentally visualise what's going on on my virtual plane when programming the acc/gyro, so I'm sure my code is working as I want it to. I've also been fiddling around with the low level code and enabled the built-in filter in the acc/gyro chip and adjusted a couple of other things as well. Man, what a difference! Now I've actually got something really useful to work with. :D I'm pretty much able to get a reliable measurement/summation of the velocity when traveling in a fairly straight line without using the gyro. Going to go for a ride most of the morning tomorrow and record a lot of drives so that i have some realistic data to work with. This should ensure that what I'm coding will also work when deployed on a real vehicle. :)

Still battling with the accelerometer and gyro project. It takes a ridiculous amount of time to get it finetuned. Infact, I'm having to rethink my whole strategy on how to calculate the measurements, so I'm in the process of reading up on linear algebra to make sure I've covered all bases.