A lot of people that I talk to about this gimbal think that I should be using servos for the motors. The main reasons seem to be that servos are easy to use and have built-in encoders (in the form of a potentiometer). However, these two items don't really cover the requirements for the project. The requirements is the starting point for identifying a motor selection.

Requirements:

1) Evenly distributed - If the distance between steps of the targeting is not evenly distributed the 3D lidar image will be stretched and skewed. This will likely prevent multiple shots from being registered together into one larger shot. A high quality encoder should not have this problem, but a low quality encoder might.

2) Highly Repeatable - If there are variations in targeting from one run to the other we will have issues with registering shots together.

3) >500 targets per second ( >1200 preferred ) - The device has to move fast enough that we can capture an entire sphere with 0.5 degree resolution (0.25 preferred)

4) Better than 0.5 degree resolution.

5) 360 degree rotation.

6) Low amounts of bounce after moving a step to next target. - Waiting for bounce to dampen requires time and slows down the process. Bounce also increases error and uncertainty. This is a downside for steppers which have some vibration during normal running.

7) Known direction with respect to earth. - note that no motor can give this information on it's own. A 6dof compass is required for determining orientation relative to earth. (And technically, you can use registration in post processing to overcome the orientation issues).

My reasoning:

The reason that I selected stepper motors was primarily because of my first requirement that my image be evenly distributed and my second requirement that the targeting be highly repeatable. Also, steppers are able to meet all of my other requirements. Also, I can always switch to servos in a later phase of the project. I'm testing this project at every step of the process. And steppers are cheap.