The automation of recurrent tasks and force feedback are complex problems in medical robotics. We present a novel approach that extends human-machine skill-transfer by a scaffolding framework. It assumes a consolidated working environment for both, the trainee and the trainer. The trainer provides hints and cues in a basic structure which is already understood by the learner. In this work, the scaffolding is constituted by abstract patterns, which facilitate the structuring and segmentation of information during “Learning by Demonstration” (LbD). With this concept, the concrete example of knot-tying for suturing is exemplified and evaluated. During the evaluation, most problems and failures arose due to intrinsic system imprecisions of the medical robot system. These inaccuracies were then improved by the visual guidance of the surgical instruments. While the benefits of force feedback in telesurgery has already been demonstrated and measured forces are also used during task learning, the transmission of signals between the operator console and the robot system over long-distances or across-network remote connections is still a challenge due to time-delay. Especially during incision processes with a scalpel into tissue, a delayed force feedback yields to an unpredictable force perception at the operator-side and can harm the tissue which the robot is interacting with. We propose a XFEM-based incision force prediction algorithm that simulates the incision contact-forces in real-time and compensates the delayed force sensor readings. A realistic 4-arm system for minimally invasive robotic heart surgery is used as a platform for the research.

中文翻译：

---

远程微创手术–医疗机器人中的触觉反馈和选择性自动化

周期性任务的自动化和力反馈是医疗机器人技术中的复杂问题。我们提出了一种新颖的方法，通过脚手架框架扩展了人机技能的转移。它假定受训者和培训者的工作环境都是统一的。培训师以学习者已经理解的基本结构提供提示和提示。在这项工作中，脚手架是由抽象模式构成的，这有助于在“通过示威学习”（LbD）期间进行信息的结构化和分段。利用该概念，示例和评估了用于打结的打结的具体示例。在评估过程中，大多数问题和故障都是由于医疗机器人系统的固有系统不精确性引起的。然后通过手术器械的视觉指导来改善这些不准确性。尽管远程手术中使用力反馈的好处已经得到证明，并且在任务学习过程中也使用了测得的力，但是由于以下原因，操作员控制台和机器人系统之间的长距离或跨网络远程连接的信号传输仍然是一个挑战。时间延迟。尤其是在用手术刀切入组织的切口过程中，延迟的力反馈会在操作员侧产生无法预测的力，并可能损坏与机器人进行交互的组织。我们提出了一种基于XFEM的切入力预测算法，该算法可实时模拟切入接触力并补偿延迟的力传感器读数。