A Design Framework for Long-Reach Manipulators in Confined Spaces Using Task- Based Kinematic Optimization and Surface Clustering
Industrial applications involving hazardous materials often require inspection within confined interiors, posing significant engineering challenges due to restricted access and complex geometries. Long-reach mechatronic systems are essential for these tasks, yet their kinematic design is typically ad hoc. This paper presents a task-driven design synthesis framework to address this gap, focusing on the optimization of manipulators for full-surface inspection inside the vacuum vessel of a nuclear fusion reactor. To facilitate 3D surface service for use in the kinematic optimization pipeline, a methodology is introduced that employs unsupervised surface clustering of CAD-derived geometry. This clustering extracts a compact set of reachability targets while maintaining coverage. Additionally, a lightweight method for 2D plane projection of the clusters enables fast collision pre-checks. Using this reduced task set, a multi-objective NSGA-II simultaneously optimizes kinematic type and dimensions, minimizing total link length and static joint torques, while ensuring position-and-orientation reachability, joint limits, and collision constraints. A representative vessel case study demonstrates that the framework achieves collision-free, full-surface coverage through a narrow access port and identifies consistent designs with fewer test points. Surface clustering-based reduction lowers computational costs by up to 60% relative to finer clustering, while maintaining task-space coverage quality. The results demonstrate a systematic and computationally efficient approach to the early-stage design of long-reach manipulators for access-limited environments. Although developed for fusion-vessel inspection, the approach is versatile and can adapt to changes in geometry, constraints, and inspection requirements.