Tokamak O&M Robot Dynamic Control by Using Optimized FABRIK and Spatial Partition Strategies in Digital Twin System

Multijoint operational robots are essential for routine maintenance tasks in a tokamak. Due to the high cost and complex internal structure of the toroidal chamber, the remote operations & maintenance (O&M) robot inside the chamber needs to have efficient real-time kinematic solving, full-body obstacle avoidance, and multimode operation switching capabilities to meet the requirements for motion accuracy and safety. In this article, we propose an O&M robot kinematics solving method based on spatial partitioning drive and the perception potential field FABRIK algorithm. This method enables real-time forward and inverse kinematic calculations and full-body obstacle avoidance for the twin O&M robot. Initially, the method involves dividing the internal cavity into equidistant sectors and planning paths to control the robot for fast and stable large-scale spatial transformations within the cavity. Upon reaching the target sector, the Perception Potential Field FABRIK algorithm is employed to perform small-scale position transformations with obstacle avoidance for the end-effector. Additionally, a corresponding digital twin system was developed to achieve virtual-real interaction and status feedback for the maintenance robot. Finally, two experimental studies were conducted: a fixed trajectory tracking experiment, and a dynamic obstacle avoidance and grasping experiment. The experimental results demonstrate that the proposed method can stably control the robot's end-effector and avoid obstacles in real-time, suggesting potential applications in other complex task control and maintenance scenarios in the future.