A techniques for improved guiding of hyper-redundant manipulator probes into a constrained space which make use of the known characteristics of the space into which the probe is being inserted to increase the efficiency of the computation of the path of the probe. Embodiments of the invention achieve this through an optimisation function which determined a new orientation which minimises the deviation between each of: a) the point on the probe where the probe starts to follow a defined curve within the constrained space and a predetermined initial point, and b) the distal end of the probe and said defined curve.