Intelligent systems for collaborative remanufacturing. AI, motion modeling, and optimization support repairability, sustainable disassembly, and human–robot collaboration.
TOPIC 01 / ROBOTICS
Intelligent Robotics for Sustainable Disassembly
My research applies AI to improve repairability and robotic disassembly in remanufacturing systems as
part of a circular economy.
Methods & contributions
Robotic Disassembly Planning: Designed a scoring framework to evaluate robotic
feasibility based on component weight, size, shape, accessibility, and positioning. Used MaxViT for
visual estimation and to recommend human–robot collaboration levels.
Repairability Prediction: Developed supervised and unsupervised learning models
using teardown images to score the repairability of electronic products. Techniques include image
classification (ConvNeXt, GoogLeNet, ResNet50, VGG16) and clustering (ORB + k-means) to uncover design patterns.
These tools support more efficient and scalable remanufacturing decisions.
Intelligent Risk Analysis and Mitigation in Robotics
This research focuses on enhancing operator safety and risk awareness for the applications of
collaborative robotic disassembly tasks.
Methods & contributions
Motion Uncertainty Modeling: Developed deep learning models (LSTM-Bagging,
GRU-Bagging, BNN-MCD) to predict ranges of hand motion trajectories using IMU data from desktop
disassembly tasks.
Vision-Based Motion Forecasting: Designed a hybrid system that combines ConvLSTM
(for future-frame generation) and YOLO (for real-time hand detection) on projected human movement.
These tools support intelligent risk mitigation in remanufacturing environments.
This research develops a decision-making framework to optimize disassembly task allocation between human
operators and robots, supporting sustainable remanufacturing.
Methods & contributions
Multi-Attribute Utility Modeling: Introduced a utility-based model that considers
disassembly cost, task complexity, and safety, while incorporating uncertainty using Beta
distributions.
Intelligent Task Assignment: Tasks are allocated based on disassembleability,
safety index, and cost profile, identifying optimal human–robot collaboration strategies.
