Sensor Calibration with Unknown （传感器校准与未知）.pdf

2013 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) November 3-7, 2013. Tokyo, Japan Sensor Calibration with Unknown Correspondence: Solving AX=XB Using Euclidean-Group Invariants Martin Kendal Ackerman , Alexis Cheng , Bernard Shiffman , Emad Boctor , Gregory Chirikjian Abstract— The sensor calibration problem proposed methods are valuable to any incarnation of must often be solved in image guided therapy systems, the problem, but we motivate our study by such as those used in robotic surgical procedures. In this problems in ultrasound sensor calibration, and illus- problem, , , and are homogeneous transformations with and acquired from sensor measurements and trate the efficacy of our approach on simulated data. being the unknown. It has been known for decades The remainder of this section is devoted to present- that this problem is solvable for when a set of exactly ing the motivating application, reviewing the literature, measured ’s and ’s, in a priori correspondence, is given. and establishing notation. Section II-A presents an algo- However, in practical problems, the data streams containing rithm using two of the well known screw theory invari- the ’ and ’s will be asynchronous and may contain gaps (i.e., the correspondence is unknown, or does not exist, ants that corrects for simple uniform shifts. Section II-B for the sensor measurements) and temporal registration is presents an algorithm using all four of the invariants required. For the problem, an exact solution can that corrects for nonuniform shifts and gaps. be found when four independent i