This paper introduces a methodology for controlling parallel robots in case they are used as a kind of specialized fixture to expand the technological capabilities of machines. The parallel robot is mounted on the workbench to extend the number of degrees of freedom. However, there are always measurable kinematic errors of the workbench which will be eliminated by the robot’s motion. The actual working motion of the robot is then still performed by its active joints. Therefore, the displacement of each movable joint is now decided by two sources, one is due to the error compensation motion of the workbench, the other is the required work movement. According to the superposition principle, these two motions are combined into a single displacement characteristic curve to control the robot. The base exchange technique to determine the error compensation motion of the workbench, the technique of solving the inverse kinematics problem by the generalized reduced gradient (GRG) method, and the principle of joint motion combination are then introduced in detail in the paper. Finally, an example with the hexapod is presented. The obtained results, which use the robot itself to generate error-compensated movements of the workbench by means of the base exchange technique, will open up the possibility of intervening in hybrid machine systems to ensure the desired forming accuracy without no hardware intervention required.