基于医生心理模型的骨科机器人直觉化交互机制与方法研究

Intuitive interaction is the goal of all surgical robot to be achieved. However, the surgeon's intention expressions and the complete capability of the intuition are not solved. In this study, we will firstly try to find out the mechanisms of intuitive interaction after being investigated of the surgeon's mental models of HRI. Secondly, we will explore the laws of key characteristics that influence the effectiveness of intuitive HRI. i.e. the laws of motion of surgical tooltip during operations under simulated robotic surgery environments by analyzing the data and process recorded from digital video recorder and motion analysis system after using work study method, and the laws of guidance force exerted by surgeon by force/torque measurement system, as well as the laws of surgeon's holding contact forces during operations will be investigated by exploiting the data from contact force measuring system. Thirdly, we will work out the methods of intuitive interaction based on admittance control by using the laws of key characteristics of intuitive interaction. That is, an effective scale factors formula of admittance control shall be obtained by using least squares method based on these results. The method of a gradient analysis to segment the areas of forbidden area virtual fixtures (FRVFs) will be proposed according to the laws of speed changing rate of the holding force. And the method of data envelope analysis to segment the areas of guidance virtual fixtures (GVFs) will be proposed according to the surface of data envelope from motion process of the surgical tooltip. Then the whole admittance control formula can be deduced for intuitive HRI. An experimental platform will be established to validate the whole proposals in the end. The aim of our study is to provide the theoretical guide to fulfill the intuitive physical HRI in orthopaedic surgery from the perspective of surgeon's ergonomics, and to provide key technologies for robot research and development during commercialized robot promotion.

针对当前外科机器人直觉化交互研究中医生意图的表达及直觉化交互能力的实现问题，从医生的心理模型分析出发，揭示骨科机器人手术过程中的直觉化交互机制。并发掘该机制中关键交互特征参数的变化及规律：通过视频录像、运动捕捉、引导力检测等手段记录医生作业的运动过程和医生施加的引导力变化过程，结合工作分析技术，揭示医生技能作业的运动特征和引导力特征规律；通过对医生作业过程中的握力特征进行检测和分析，揭示握力触觉变化规律。进而将这些特征参数规律应用于基于导纳控制的直觉化交互方法实现中：拟合出基于最小二乘方法的导纳增益参数表达形式；提出了基于握力梯度的禁止区域虚拟夹具边界划分方法和基于运动轨迹数据包络面的引导区域虚拟夹具边界划分方法，以此优化了虚拟夹具参数表达；最后进行直觉化交互能力的实验验证。为解决直觉化人-机器人交互提供理论指导，为未来骨科机器人面向临床提供关键技术支撑，因而具备理论的和工程的研究意义。

针对当前外科手术机器人直觉化交互研究中医生意图表达及人机直觉化交互能力问题，本研究基于人体运动过程最小加加速度理论模型，构建出医生常规手术中手持手术工具移动过程的运动特征模型。并将此模型作为直觉化的人机交互意图表达，借助模糊模型参考学习控制理论，从而构建基于可变导纳控制的人机直觉化交互方法，验证结果表明，该方法能够根据个人特性，快速训练到其习惯的运动特征上，从而较好实现了直觉化交互能力。另外，针对手术区域边界安全问题，借助常量导纳控制技术，降低接近禁止区域边界法向力的方法，实现柔顺控制的同时，使机器人的操作被限制在禁止区域以内，实验结果表明，该方法收敛性能够得到保障，无论医生引导力如何变化，其速度在向禁止区域的收敛仍然是平缓的，因而能保证手术过程人机交互的柔顺性。本研究的实施，为人机协作模式骨科手术机器人面向临床提供关键技术支撑。同时，为人机物理交互环境下的人机直觉化交互提供有价值的借鉴，对外骨骼、康复等与人体直接接触的交互能力提供直接的方法。