Related Experiment Video
Updated: Oct 3, 2025
10:09
Operation of the Collaborative Composite Manufacturing CCM System
Published on: October 1, 2019
6.7K
Robotic Manipulation Planning for Automatic Peeling of Glass Substrate Based on Online Learning Model Predictive Path
Liwei Hou1, Hengsheng Wang1,2, Haoran Zou1
1College of Mechanical and Electrical Engineering, Central South University, Changsha 410083, China.
Sensors (Basel, Switzerland)
|February 15, 2022
Summary
This study introduces an Online Learning Model Predictive Path Integral (OL-MPPI) algorithm for autonomous robotic peeling of glass substrates. The method uses deep learning and online tuning to ensure safe and efficient manipulation.
Area of Science:
- Robotics
- Artificial Intelligence
- Materials Science
Background:
- Autonomous planning for robotic contact-rich manipulation is complex.
- Automatic peeling of LCD glass substrates demands high safety standards.
Purpose of the Study:
- To develop an automated system for safe and efficient glass substrate peeling.
- To enable robots to perform delicate contact-rich manipulation tasks.
Main Methods:
- A system model was established using data and deep learning for pretraining.
- An online learning algorithm tuned the model with real-time experimental data.
- The Online Learning Model Predictive Path Integral (OL-MPPI) algorithm was proposed for optimal robot trajectory planning.
Main Results:
- The OL-MPPI algorithm demonstrated effective trajectory planning for robot manipulation.
- The system successfully performed automatic glass substrate peeling tasks in experiments.
- The approach balances simulation-based pretraining with real-time adaptive learning.
Conclusions:
- The developed OL-MPPI algorithm provides a robust solution for autonomous robotic contact-rich manipulation.
- This method enhances safety and efficiency in tasks like automated glass substrate peeling.
- The combination of deep learning and online learning addresses real-world process uncertainties.
