In a robotic endoscope system, the orientation of a captured camera view at a distal tip of a robotic endoscope and displayed on a screen viewable by an operator of the endoscope is automatically maintained at a roll orientation associated with a setpoint so as not to disorient the operator as the endoscope is moved, flexed and its tip turned in different orientations. A processor generates a current commanded state of the tip from operator input and modifies it to maintain the setpoint roll orientation. To generate the modified current commanded state, the current commanded roll position and velocity are constrained to be a modified current commanded roll position and velocity that have been modified according to a roll angular adjustment indicated by a prior process period commanded state of the tip and the setpoint. The processor then commands the robotic endoscope to be driven to the modified commanded state.