Cable-driven parallel robots(CDPRs)use cables instead of the rigid limbs of traditional parallel robots,thus processing a large potential workspace,easy to assemble and disassemble characteristics,and with application...Cable-driven parallel robots(CDPRs)use cables instead of the rigid limbs of traditional parallel robots,thus processing a large potential workspace,easy to assemble and disassemble characteristics,and with applications in numerous fields.However,owing to the influence of cable flexibility and nonlinear friction,model uncertainties are di cult to eliminate from the control design.Hence,in this study,the model uncertainties of CDPRs are first analyzed based on a brief introduction to related research.Control strategies for CDPRs with model uncertainties are then reviewed.The advantages and disadvantages of several control strategies for CDPRS are discussed through traditional control strategies with kinematic and dynamic uncertainties.Compared with these traditional control strategies,deep reinforcement learning and model predictive control have received widespread attention in recent years owing to their model independence and recursive feasibility with constraint limits.A comprehensive review and brief analysis of current advances in these two control strategies for CDPRs with model uncertainties are presented,concluding with discussions regarding development directions.展开更多
基金Supported by National Natural Science Foundation of China(Grant No.51525504)。
文摘Cable-driven parallel robots(CDPRs)use cables instead of the rigid limbs of traditional parallel robots,thus processing a large potential workspace,easy to assemble and disassemble characteristics,and with applications in numerous fields.However,owing to the influence of cable flexibility and nonlinear friction,model uncertainties are di cult to eliminate from the control design.Hence,in this study,the model uncertainties of CDPRs are first analyzed based on a brief introduction to related research.Control strategies for CDPRs with model uncertainties are then reviewed.The advantages and disadvantages of several control strategies for CDPRS are discussed through traditional control strategies with kinematic and dynamic uncertainties.Compared with these traditional control strategies,deep reinforcement learning and model predictive control have received widespread attention in recent years owing to their model independence and recursive feasibility with constraint limits.A comprehensive review and brief analysis of current advances in these two control strategies for CDPRs with model uncertainties are presented,concluding with discussions regarding development directions.
