Abstract
Fuel debris retrieval at the bottom of the primary containment vessel (PCV) is one of the significant tasks for the decommissioning of the nuclear power plant and in particular for 1F. It is challenging for conventional manipulators to perform the retrieval process due to the presence of radiation, water leakage, and poor lighting conditions. We tackle those problems with the design and fabrication of a novel mechanical manipulator and its control and navigation algorithm. CVT (Continuous Variable Transmission)-based actuation improves the robot’s shock resistance. AI-based navigation algorithm enables semiautonomous navigation and grasping in the cluttered environment inside the PCV.
First, we investigated the shock-resistant mechanisms for the drive train and the gripper part. The drive train features CVT-VIA which utilizes the toroidal CVT for the adaptive gearing. In addition, the flexible debris gripper makes use of the spring joint instead of the conventional axial joint. The tuning of ligament allocation enables the designer to devise the spring joint with the arbitrary deformation characteristics.
Second, we developed a navigation system to overcome the obstacles. The human operator can control the robot end effector by clicking the target point from the vision input. The method features the ML-based approach to overcome the cluttered conditions.
Third, we validated the approach with a real-world experiment. The experiment was done at NARREC.
In conclusion, the decommissioning robot manipulator features the CVT-based actuation and a learning-based navigation system. Future works include the development of the whole manipulator with CVT-VIAs and the integration of the navigation system with the compliant actuators.
