From classic to learning-based algorithms:A survey of cooperative control for dual-arm systems
The cooperative control of dual-arm systems(DAS)is a technology that plays an important role in the field of robotics.It can assist human beings to perform complex and dangerous tasks in unstructured environments such as industrial production,domestic life,space and underwater.However,the aforementioned technology is a challenging topic due to difficulties in the design of controllers,which is caused by strong coupling,high nonlinearity and uncertainties.In this survey article,we first review the development of dual-arm systems,and then give an introduction of the structure,modelling,control and application of the systems.In particular,the cooperative control method of the dual-arm systems is concluded in detail:first,classic methods such as master-slave control,force/position hybrid control and impedance control during collaborative manipulation;and second,intelligent control method based on neural network and fuzzy system;third,the latest progress of data-driven methods based on reinforcement learning in robot control.Finally,the future development trends of dual-arm systems are envisioned.