Ewan Lin

Research Overview

My research connects mechanics with AI — using generative and surrogate models to accelerate robot motion solving and engineering structural optimization. Below are my focus areas and specific projects.

Focus Areas

Research Interests

Robot Inverse Kinematics & Motion Planning

Fast inverse kinematics and trajectory planning for redundant and hyper-redundant manipulators, with multi-solution generation and constraint-aware selection.

Generative Models for Engineering

Applying conditional normalizing flows and other deep generative models to physical and mechanical problems where one input admits many valid solutions.

Surrogate Modeling & Structural Optimization

Kriging and data-driven surrogates that replace expensive finite-element simulation, combined with intelligent optimization for constrained lightweight design.

Mechanics Simulation & FEA

Parametric modeling and automated simulation with Abaqus, ANSYS, and SolidWorks, including Python-driven secondary development for batch data generation.

Experience

Research Experience

  • Fast Inverse Kinematics and Trajectory Planning for an 8-DOF Redundant Manipulator · Core Researcher

    Oct 2024 – Present
    • Built a conditional normalizing-flow generative model for an 8-DOF hybrid-joint redundant manipulator that solves inverse kinematics for arbitrary end-effector poses in milliseconds and returns multiple feasible joint solutions at once.
    • Modeled the kinematics of a hybrid (revolute + prismatic) redundant arm and trained a generative IK network that produces multiple solutions per query.
    • Designed a constraint-based selection scheme to pick the optimal solution under joint limits, obstacle avoidance, and motion continuity.
    • Refined generated solutions with an improved optimization algorithm to reach industrial-grade end-effector accuracy.
    • Completed Cartesian- and joint-space trajectory planning for smooth, continuous, and efficient motion control.

  • Mechanics Analysis and Lightweight Optimization of a Double-Jib Derrick Waist Ring · Core Researcher

    Nov 2024 – Jun 2025
    • Built a global derrick model and a local roller–angle-steel contact model of the first waist ring, identified the governing cases, simplified the roller layout, and sized the pipe section with Kriging plus GWO, cutting 2.88 kg per piece.
    • Analysed nine hoisting and wind cases: Case 3 is the unbalanced-hoist extreme, Case 9 the storm-unloaded extreme; the original ring peaks at 121.80 MPa on the node-2 connecting plate.
    • Reduced rollers from 16 to 8 and simplified connecting-plate holes, moving the peak from the plate onto a load-bearing roller.
    • Fitted a Kriging surrogate of the governing-case peak stress and ran GWO over OD 30–37 mm and thickness 3–5 mm; the optimum 32.398 mm × 3 mm section saves 2.88 kg per piece.

  • Lightweight Redesign of a Power-Tower Lifting-Frame Mechanism · Structural Design & Simulation

    Jul 2025 – Present
    • Redesigned a power-tower lifting frame by replacing the conventional rope-lifting scheme with a lead-screw mechanism, achieving an integrated and lightweight structure verified by finite-element analysis.
    • Proposed an integrated redesign that replaces rope lifting with a lead-screw drive, effectively reducing the mechanism's volume and weight.
    • Completed motor selection and lifting-platform structural design following engineering steel-selection standards, with 3D modeling in SolidWorks.
    • Performed stress and displacement analysis of key load-bearing structures in ANSYS to verify that the design meets strength and stiffness requirements.