Object Manipulation with Physics-Based Models - Mehmet Dogar, University of Leeds
Paul G. Allen School via YouTube
Overview
Explore robotic object manipulation techniques in this 58-minute lecture from the Spring 2021 Robotics Colloquium. Delve into physics-based planning for cluttered scenes and grasping-based manipulation for forceful operations. Learn about motion planners that predict object movements upon robot contact, and discover strategies for robots to assist humans in tasks like drilling and cutting. Examine the geometric, force stability, and human-comfort constraints in collaborative human-robot scenarios. Gain insights from Associate Professor Mehmet Dogar of the University of Leeds as he shares his research on model predictive control, multi-object manipulation, and the challenges of accuracy in robotic systems. Conclude with open questions and a discussion on assumptions in the field of robotic manipulation.
Syllabus
Introduction
Object Manipulation
Forceful Collaboration
Physicsbased Object Manipulation
Model Predictive Control
Limitations
Accuracy
Worthless accuracy
Example tasks
Model hierarchy
Multiple objects
Course models
Acknowledgement
Human Robot Force of Collaboration
Human Comfort
Measurements
Open Questions
Conclusion
Questions
Assumptions
Taught by
Paul G. Allen School