Modified Dynamic Movement Primitives With Control: A Novel Trajectory Planning Framework for Dual-Arm Asynchronous Tasks

In industrial environments, such as pharmaceutical applications, dual-arm asynchronous cooperation is essential for precise temporal and spatial coordination, particularly in tasks like handling liquid. Using dynamic movement primitives (DMPs), we establish separate DMPs models for each arm, with synchronization achieved through a shared canonical system. However, the original DMPs framework often struggles to accurately reproduce trajectory shapes and achieve endpoint precision, which is crucial in delicate tasks like liquid handling. To address these limitations, we propose a novel framework that integrates DMPs with model predictive control (MPC) to enhance the precision and reliability of dual-arm asynchronous tasks. Our approach introduces an MPC-generated coupling term within the DMPs formulation, continuously optimizing both trajectory shape and endpoint accuracy. The proposed method was validated in six numerical experiments, demonstrating an average improvement of 81.37% in dual-arm endpoint position error control efficiency compared to the original DMPs approaches. Finally, real-world validation using two ABB GoFa CRB 15 000 robots demonstrated the effectiveness of the framework in executing precise liquid handling tasks in industrial settings.