Design and Control of Robotic Systems and Devices for Medical Applications

拟议的研究计划旨在为下一代微创医疗干预开发先进的机器人和智能系统，以最大限度地减少侵入性和创伤，同时提高安全性、效率、可靠性和成本效益。因此，有必要研究机器人辅助医疗干预的现状，例如，利用连续体机器人技术、机器学习、“智能”驱动和传感技术的最新发展，以及触觉和远程操作的进步。拟议的研究计划将通过在以下应用背景下开发各种技术和技术来解决这一需求：******同心管机器人（CTRs）的设计和实时控制框架：CTR是一种连续体机器人设备，由一组不同弯曲轮廓的同心管组成。通过相互旋转和平移这些管子，它们的曲率可以用来获得整个机器人的不同曲线轮廓，并定位和定向机器人的尖端，同时避开手术工作空间内和周围的敏感区域。CTR的尺寸限制和形状使力传感具有挑战性。灵活性和形状的变化给控制体系结构的设计以及工作空间的确定和路径规划带来了极大的复杂性。此外，CTR的灵活性使得触觉远程操作的问题特别具有挑战性。提出的框架将提供一种方法来解决诸如特定干预所施加的解剖学和其他限制、运动灵活性和工作空间要求、要使用的成像和其他传感模式、控制问题（如位置和力控制、对建模和测量不确定性的鲁棒性）以及触觉和远程操作等问题。****** (RA)MIS的手术训练和技能评估自主系统：该研究项目将建立在最近在受控环境中进行的初步工作的基础上，这些工作已经证明了在机器人辅助微创手术（RAMIS）中引入自主缝合任务的可能性。提出的研究将探索如何使用机器人平台和各种传感模式来分析各种机器人辅助或传统微创手术任务的性能。然后，机器学习方法可以使用得到的数据来训练手术机器人系统，使其成为“智能虚拟专家（IVE）”，从而可以代替实际的专家外科医生进行离线（RA）MIS培训。将研究诸如深度神经网络、循环神经网络、强化学习及其变体等方法的应用。＊＊＊＊＊

The proposed research program is aimed at developing advanced robotic and intelligent systems for the next generation of minimally invasive medical interventions that will minimize invasiveness and trauma while enhancing safety, efficiency, reliability and cost-effectiveness. There is therefore a need to examine the current state of robotics-assisted medical interventions, for example by using recent developments in continuum robotics, machine learning, and “smart” actuation and sensing technologies, coupled with advances in haptics and teleoperation. The proposed research program will address this need through the development of various technologies and techniques in the context of the following applications:******Framework for the Design and Real-Time Control of Concentric-Tube Robots (CTRs): A CTR is a type of continuum robotic device consisting of a set of thin concentric tubes, shaped in different curved profiles. By rotating and translating the tubes with respect to each other, their curvatures can be used to obtain different curved profiles for the overall robot and to position and orient the robot's tip while avoiding sensitive regions in and around the surgical workspace. The CTR's size constraints and shape make force sensing challenging. The flexibility and changing shape introduce significant complexity in the design of a control architecture as well as in workspace determination and path planning. Furthermore, flexibility in the CTR makes the problem of haptics-enabled teleoperation particularly challenging. The proposed framework will provide a methodology to address issues such as anatomical and other constraints imposed by particular interventions, the kinematic dexterity and workspace requirements, the imaging and other sensing modalities to be used, control issues (such as position and force control, robustness to modeling and measurement uncertainty), and haptics and teleoperation.******Autonomous Systems for Surgical Training and Skills Assessment for (RA)MIS: This program of research will build on recent preliminary work conducted in controlled environments that has demonstrated the possibility of introducing autonomy in suturing tasks during Robotics-Assisted Minimally Invasive Surgery (RAMIS). The proposed research will explore how a robotic platform and a variety of sensing modalities can be used to analyze the performance of various robotics-assisted or conventional minimally invasive surgery tasks by experts. Machine learning approaches can then use the resulting data to train a surgical robotic system to serve as an “Intelligent Virtual Expert (IVE)” so that it can be used in place of an actual expert surgeon for offline (RA)MIS training. Application of approaches such as Deep Neural Networks, Recurrent Neural Networks, Reinforcement Learning and their variations will be investigated. *****