Design Optimization of a 3-PUU Parallel Machine for Friction Stir Welding Robots

Friction Stir Welding (FSW) has recently emerged as an effective solid-state joining technique for welding high-strength aluminum alloys and light metals, particularly in the railway industry. However, a limitation of using FSW in this industry is the restricted workspace available. To address this issue, a 3-PUU (Prismatic and Universal Joints) parallel machine with 3-DoF (Degrees of Freedom) translational movement is utilized to support the FSW robot’s tool head, allowing the robot to operate over a larger horizontal workspace. This paper focuses on the design and optimization of a 3-PUU parallel machine to achieve an optimal configuration based on performance indices, specifically the Local Conditioning Index (LCI) and the dynamic isotropy index ( $d_{1}$ ). Initially, a design space atlas was created, mapping both LCI and $d_{1}$ were identified. A Genetic Algorithm (GA) was then used to determine a single optimal design within this region. A cuboid workspace was defined for this optimal design, and two test trajectories were generated for the robot to follow. A prototype of the optimal design was built, and a series of experiments were conducted to verify and validate the kinematic and dynamic models within the defined cuboid workspace, which provides foundations for further industrial applications of the proposed machine in FSW.