Grasping is a fundamental skill that robots will need when they start working in everyday environments. But there are very large challenges with grasping - where should the robot put its fingers? How hard should it squeeze the object? What happens if it slips? What if the robot has never seen the object before? These are tasks that humans can handle without a second thought, but they still vex even the most advanced robots.
In our lab we are working on ways to make robots grasp objects in a way that is safe for the object and efficient enough for use in the home. Current research projects include robust slip detection and force regulation, multifingered dextrous manipulation, and intelligent software architectures for grasping under uncertainty.