I imagine that engineers have to decide where an inertial measurement unit (or just an accelerometer, gyro, etc.) is placed in a rocket or spacecraft. I'd like to know how are these positions decided? Do they put them in zones of least vibrations for the smoothest signal, etc.? Particularly for sensor fusion, how would one decide on sensor location for the best attitude estimation of a rocket? Thanks for your input!
For spacecraft it's "where they fit and can be oriented properly". Harnessing constraints are somewhat considered (how do you get wires to it) and thermal constraints might be pretty important (ir telescopes need the detector to be very very cold so anything that produces heat is as far away as possible), but in general for satellites all motion is pretty slow so vibration isn't much of an issue. I can't talk about rocket design.
See NASA Technical Note D-5869. This document includes a discussion about choosing the location of the navigation instruments of the Saturn V rocket. See the figures on pages 11 and 13 and nearby text (page 12).
Several considerations were mentioned, including the effects of bending modes, and other environmental factors of possible locations, length of cabling needed, and so on.
The bending mode shapes show the magnitude of the amplification of lateral motion. You would prefer not to place motion sensors at a position that bends (a lot) at a similar frequency to control motions of the rocket, as the bending could feed back into the control loop and lead to instability.
It depends upon the mission. Ideally you try to put them as close to the center of mass as you can so that you don't have that offset between the IMU and the center of mass to worry about, but these days with everything modeled up to the millimeter in CAD, even that isn't that important any more. In practice, unless there is a mission driver, you put it where it makes sense for practical reasons, like to balance the mass distribution.