Fabel Stephan, Baek Kyungim, Berkelman Peter
Human-Robot Interaction Laboratory, Department of Information and Computer Sciences, University of Hawaii at Mānoa, 2540 Dole St., Honolulu, HI 96822, USA.
Annu Int Conf IEEE Eng Med Biol Soc. 2010;2010:971-4. doi: 10.1109/IEMBS.2010.5627598.
The Compact Lightweight Surgery Robot from the University of Hawaii includes two teleoperated instruments and one endoscope manipulator which act in accord to perform assisted interventional medicine. The relative positions and orientations of the robotic instruments and endoscope must be known to the teleoperation system so that the directions of the instrument motions can be controlled to correspond closely to the directions of the motions of the master manipulators, as seen by the the endoscope and displayed to the surgeon. If the manipulator bases are mounted in known locations and all manipulator joint variables are known, then the necessary coordinate transformations between the master and slave manipulators can be easily computed. The versatility and ease of use of the system can be increased, however, by allowing the endoscope or instrument manipulator bases to be moved to arbitrary positions and orientations without reinitializing each manipulator or remeasuring their relative positions. The aim of this work is to find the pose of the instrument end effectors using the video image from the endoscope camera. The P3P pose estimation algorithm is used with a Levenberg-Marquardt optimization to ensure convergence. The correct transformations between the master and slave coordinate frames can then be calculated and updated when the bases of the endoscope or instrument manipulators are moved to new, unknown, positions at any time before or during surgical procedures.
夏威夷大学研发的紧凑型轻量化手术机器人包含两个远程操作器械和一个内窥镜操纵器，它们协同工作以实施辅助介入医学操作。远程操作系统必须了解机器人器械和内窥镜的相对位置及方位，这样才能控制器械运动方向，使其与主操纵器的运动方向紧密对应，这一对应关系通过内窥镜观察并呈现给外科医生。如果操纵器基座安装在已知位置且所有操纵器关节变量已知，那么主从操纵器之间所需的坐标变换就能轻松计算出来。然而，通过允许内窥镜或器械操纵器基座移动到任意位置和方位，而无需重新初始化每个操纵器或重新测量它们的相对位置，可以提高系统的通用性和易用性。这项工作的目的是利用内窥镜摄像头的视频图像来确定器械末端执行器的位姿。采用带Levenberg-Marquardt优化的P3P位姿估计算法以确保收敛。这样，在内窥镜或器械操纵器基座在手术前或手术过程中的任何时间移动到新的未知位置时，就能计算并更新主从坐标系之间的正确变换。
