I've used "A General Construction Scheme for Unit Quaternion Curveswith Simple High Order Derivatives" in the past, and while not perfect it was generally good

I've used "A General Construction Scheme for Unit Quaternion Curveswith Simple High Order Derivatives" in the past, and while not perfect it was generally good enough and fairly easy to implement. Basically it extends Hermite splines to Quaternion splines using the Lie group operations.