We present an orientation adaptive controller to compensate for the effects of highly constrained environments on continuum manipulator actuation. A transformation matrix updated using optimal estimation techniques from optical flow measurements captured by the distal camera is composed with any Jacobian estimation or kinematic model to compensate for these effects. By utilizing domain knowledge to define the structure of this matrix, fewer parameters need to be estimated and a stable controller can be guaranteed. The algorithm is tested on a custom robotic catheter and convergence is shown both empirically and theoretically.

Model-free Visual Control for Continuum Robot Manipulators via Orientation Adaptation