Synergy control of a five-fingered musculoskeletal robot hand based on coordination of agonist-antagonist muscle

在医疗和老年护理行业中，能够与人共存的机器人必须安全灵活，因为它们直接与人类一起工作。气动执行器对于实现这一目标是有用的，因为它们是具有自然顺应性的轻质单元。我们的研究重点是类似人类的机器人关节配备气动执行器，以模仿人体肌肉骨骼结构的机制和生物启发的控制，用于驱动的五指肌肉骨骼机器人手，每个关节都有一对拮抗肌肉。(1)我们从理论和实验两方面研究了所提出的类人关节模型的有效性，（2）测量和统计分析了人手的抓取姿态和力分布，（3）开发了一个无传感器的关节角度/接触力估计系统，最后（4）通过实现肌肉骨骼机器人手演示了一种生物启发的控制算法。

Robots that can coexist with people in medical and aging care industries must be safe and flexible because they work directly with humans. Pneumatic actuators are useful for achieving this goal because they are lightweight units with natural compliance. Our research focuses on human-like robotic joints fitted with pneumatic actuators to mimic the mechanism of the human musculoskeletal structure and bio-inspired control for a pneumatically driven five-fingered musculoskeletal robot hand, which has an antagonistic muscle pair for each joint. (1) We have investigated the validity of the proposed human-like joint model both theoretically and experimentally, (2) measured and statistically analyzed grasping posture and force distribution of the human hand, (3) developed a sensor-less joint angle / contact force estimation system, and finally (4) demonstrated a biologically inspired control algorithm by implementing the musculoskeletal robot hand.