Reaching for Dexterous Manipulation

I review three papers in the area of dexterous manipulation by machines. These papers focus on different stages of the process of grasping an object. The first describes a method for generating a suitable hand pose to grasp an object at a given set of contact points. The second describes a method for adapting an example manipulation task from a given object to a new object. The third describes a control-theoretic method for iteratively improving the quality of a given grasp of an object of unknown geometry and using only tactile sensory information. These three papers are good examples of mainstream work in dexterous manipulation. Following a brief review of the essential idea presented by each of these three papers, I consider the relationships between each of these approaches, focusing on the larger issues that loom in the differences between them. Finally, I propose some new ideas to try to open the door to dexterous manipulation in more ordinary, unstructured human environments.