刚柔耦合外肢体机器人运动学与人机协作控制研究

Kinematic Analysis of Rigid-Flexible Coupled Supernumerary Robotic Limbs and Human-Robot Collaborative Control Research

摘要:

针对复杂集总干扰下刚柔耦合绳驱动外肢体机器人控制精度低、人机运动不相容及协作效能差等问题，研究了一种刚柔耦合外肢体机器人运动学及人机协作控制方法。基于D-H法和关节耦合机理构建了绳驱动外肢体机器人主动解耦运动学模型，设计了一种基于惯性传感器的人体上肢运动捕捉系统，实现对上肢运动数据的动态采集，并建立基于神经网络的人体上肢运动识别模型，提出了一种基于耦合运动学模型和上肢运动意图的人机协作控制策略，最后搭建了单臂绳驱动外肢体机器人实验样机，对所建的运动学模型和人机协作控制策略进行验证。结果表明，8组不同形态下人体运动意图识别准确率高达99.75%，外肢体机器人末端位置偏差平均值为4.8mm，位置偏差最大值为6.5mm，验证了所建模型及人机协作控制策略的可行性和正确性，确保了高性能的安全控制和人机协作效能。

Abstract:

Aiming at the problems of low control accuracy, human-machine motion incompatibility and poor collaborative effectiveness of rigid-flexible coupled rope-driven supernumerary robotic limbs under complex aggregate interference, an inertial sensor-based human upper limb motion intention recognition and collaborative control method was studied. Based on the D-H method and joint coupling mechanism, an active decoupling kinematic model of rope-driven supernumerary robotic limbs was constructed, an inertial sensor-based human upper limb motion capture system was designed to realize the dynamic collection of upper limb motion data, a neural network-based human upper limb motion recognition model was established, and a human-robot collaborative control strategy based on the coupled kinematic model and the upper limb motion intention was proposed. Finally, a single-arm rope-driven experimental prototype of supernumerary robotic limbs was built to validate the proposed upper limb motion recognition method and human-robot collaborative control strategy. The results showed that the accuracy of human motion intention recognition in eight groups of different morphologies was as high as 99.75%, while the mean value of the end position deviation of the supernumerary robotic limbs was 4.8mm, and the maximum value of the position deviation was 6.5mm, which verified the feasibility and correctness of the proposed method of human motion recognition and human-robot collaborative control strategy, and ensured high-performance safety control and human-robot collaborative effectiveness.