Nonlinear model predictive control (NMPC) based trajectory tracking on EAST Articulated Maintenance Arm (EAMA)

Abstract

• A nonlinear model predictive control based trajectory tracking controller is formulated. • Kinematics model and collision avoidance constraints are both automatically satisfied. • A novel two-step recursive algorithm is developed for efficient solving of NMPC. EAST Articulated Maintenance Arm (EAMA) is a 8-DOF redundant articulated serial manipulator utilized to conduct maintenance tasks in EAST (Experimental Advanced Superconducting Tokamak). Due to the redundancy and total length (8.7395 m) of manipulator and the narrow space in the CASK and the curved vacuum vessel(VV), it is difficult to perform online collision-free end effector trajectory tracking on EAMA. Current solution is to record a feasible trajectory offline by the operator after thousands of trials and errors as the reference for EAMA to track, which is time-consuming, inefficient, not friendly to operators, absence of resistance to external disturbances and difficult to be combined with other maintenance tasks like visual servoing. This paper proposes a Nonlinear Model Predictive Control (NMPC) based online trajectory tracking method, which is easy to use and flexible. The trajectory tracking problem is elegantly formulated as an optimization problem to minimize the end effector's tracking error while satisfying nonlinear constraints consisting of boundaries of state, output and control input, kinematic model of EAMA and collision avoidance constraints. Besides, a two-step recursive solver is developed to speed up the solving of NMPC problem to guarantee real-time control. The effectiveness and good performance of this method are demonstrated by an inspection simulation where the end effector tracks a semicircular trajectory.