Service satellite are used in a variety of applications, justified by scientific, economic, strategic and social benefits. The perspective of working with space equipment is largely based on the assumption that there are robotic mechanisms capable of handling multiple payloads. Normally, robot path planning involves determining the inverse kinematics configuration, since the workspace is expressed in Cartesian coordinates and the robot control is performed through joint angles. The present study discusses strategies for calculating the inverse kinematics of a space serial manipulator. The results of a classical industrial kinematics model and a multibody dynamics formulation are compared in the context of space operation. A Neural Network model is considered to quantify the uncertainty about the end-effector positioning and joint angle values. It was found that the Neural Network is a robust predictor that, together with a Nonlinear Programming strategy, represents an effective methodology for inverse kinematics calculations. This result contributes to the establishment of algorithms that will increase the operational autonomy of space equipment in different scenarios of orbital operation.