Fully automatic needle calibration for robotic-assisted puncture on small animals

In vision-based medical applications involving tools attached to a robotic arm, it is essential to be able to accurately localize these tools in the robot end-effector frame. Indeed, inaccurate tool-calibration has a direct impact on the accuracy of the task. In this paper, we introduce a versatile and robust calibration technique to estimate the relative position between the robot end-effector and a needle, independently of the needle length and thickness. Firstly, we propose a new method to calibrate the needle pose using a virtual frame created from pre-defined needle motions. Secondly, we compare the obtained results to a gold standard and show that the proposed calibration method provides an error within 0.35 mm.