At the workshop, we will address a number of questions w.r.t. grasping, which should lead us to understanding the difference between model-based and model-free grasping. The following basic questions are a starting point for that:
- 2010-03-27: What are the most common grasp failure modes? What are the causes and how could they be prevented?
- 2010-04-03: What information about the object and environment can we safely assume is available?
- 2010-04-12: What are the mechanisms behind human grasp acquisition that make it so successful? Do we implement impedance control? If so, what are the inputs? How is sensing exploited? Can robots be designed to mimic this exploitation?
- 2010-04-21: (Adult) Human grasping is mostly based on learned object models. What consequences do object model details have on grasp functionality?
- 2010-04-27: Can we break the problem into manageable chunks? Should we?
- 2010-05-06: How do uncertainties affect current grasping approaches?
- 2010-05-14: How do we achieve robustness to uncertainties and control errors?
- Question of the week: The dominant model-based approach is quasi-static: choose finger locations, place fingers there without disturbing the object, then squeeze. When does this approach fail? Why?
Please take your time to give these issues a thought and enter the discussion blog.