This paper presents an open-architecture of CNC system and mirror milling technology for a new-type 5-axis hybrid robot named TriMule. The CNC system with dual CPUs is developed first to achieve human-computer interaction and motion control. Then, three key technologies are integrated in the system for improving the control quality, including singularity avoidance, feedforward control considering joint couplings and real-time error compensation by using externally mounted encoders. Based on these control technologies for single robot system, a collaborative machining strategy on the mirror milling system that consists of two TriMule robots is proposed to control the machining wall thickness of large thin-walled structural parts. Experiments on the TriMule robot and mirror milling system verify that the acceptable machining accuracy on the NAS test part and large thin-walled structural part can be ensured by using the developed CNC system and technologies. The root mean square of wall thickness error using the collaborative machining strategy can be 41.67% lower than the case without using the strategy.

本文介绍了一种用于新型 5 轴混合机器人 TriMule 的数控系统和镜面铣削技术的开放式体系结构。率先开发出双CPU数控系统，实现人机交互和运动控制。然后，在系统中集成了三个关键技术来提高控制质量，包括奇点避免、考虑关节耦合的前馈控制和使用外部安装的编码器的实时误差补偿。基于这些单机器人系统控制技术，提出了一种由两个 TriMule 机器人组成的镜面铣削系统的协同加工策略，用于控制大型薄壁结构件的加工壁厚。在TriMule机器人和镜面铣削系统上的实验验证了使用开发的数控系统和技术可以确保NAS测试件和大型薄壁结构件的可接受加工精度。使用协同加工策略的壁厚误差均方根可以比不使用该策略的情况降低41.67%。