Hello, I've been trying to understand exactly how TF2 works. I'm working on using sensor fusion to scale a monocular SLAM system, and I need to transform between the SLAM's artificial world frame, the camera's frame, and the frame of the accelerometer/gyro which is fixed on a board next to the camera.

I'm having a very difficult time understanding the "big picture" of how TF2 is supposed to be used. I've gone through all the tutorials, and I'm left with a lot of questions. As fair as I can tell, the buffer.transform method is the only way to actually apply coordinate transforms. However, as far as I have been able to tell, this only works with positional data.

To find the find the velocity or acceleration from raw sensor data, at a point rigidly displaced from the sensor, it seems to me that I need to take into account rotational velocity and maybe acceleration as well. I also have raw gyro data for this purpose, but it's not clear how to do this with the methods I know of in TF2. Frankly, the details of the math are a little confusing to me here as well, but I can't seem to find any good resources that discuss it. Being able to account for the location of inertial sensors on a robot seems to a very fundamental task, but so far I'm finding it surprisingly difficult to make progress with it.

Any pointers would be very helpful.