突发吞咽、呛咳引起的位移会显著增加咽喉机器人手术的安全风险，目前缺乏针对性的解决方案。在前期 “支撑喉镜专用手术机器人系统”基础上，申请人进行理论研究，创立基于仿生学的“图像识别-自动反馈-术者监控”应急安全保护模式；并以此为基础，展开功能性研究。首先进行二维内镜图像三维配准、计算机图像跟踪及识别相关算法的优化；然后采用分层强化学习算法，深度学习先验临床数据和大动物动态实验数据，实现准确自动识别不同“手术场景”下“吞咽和/或呛咳状态”作出相应机（器）械位姿调整，研发出面向术中吞咽呛咳突发事件的支撑喉镜手术机器人智能化应急安全体系。本研究将为研发更为安全的自有知识产权的喉部专用手术机器人打下基础，填补当前手术机器人智能应急安全保护系统研究的空白，实现手术机器人性能与安全性并重，造福广大患者。

Sudden swallowing and cough-induced displacement can significantly increase the safety risk of laryngoscopic robot surgery. To date there is no specific solution. On the basis of previous work on "self-retaining laryngoscope specific surgical robot system", theoretical research will be initiated to build a human physiological reflection based emergency protection mode, which is "image recognition-automatic feedback-operator monitoring". Functional research then will be conduct. Three-dimensional registration of two-dimensional endoscopic images, computer image recognition and motion tracking related algorithms will be optimized. Through deep learning of prior clinical data and animal dynamic experimental pooled data, layered reinforcement learning algorithm will be utilized to automatically identify different task subsets, which are formed by “surgical scenes” and “swallowing and/or coughing states”. By these steps, an intelligent emergency protection system for self-retaining laryngoscope surgery to deal with intraoperative swallowing and coughing emergencies will be developed. This research will lay a foundation for the development of a safer laryngeal surgery robot. It will fill the gap of current research on the intelligent emergency protection system of surgical robots, and achieve equal emphasis on high performance and safety for surgical robots, so as to benefit the patients.