Building a generic architecture for robot hand control

As various dextrous robot hands are designed and built, a major question is how to develop device-independent robot hand controllers. This would allow the low-level control problems to be separated from high level functionality. GeSAM is a generic robot hand controller that is based on a model of human prehensile function. It focuses on the relationship between geometric object primitives and the ways a hand can perform prehensile behaviors. The authors show how the relationship between object primitives and a useful set of grasp modes can be learned by an adaptive neural network. By adding training points as necessary, system performance can be improved, avoiding the tedious job of computing every relationship by hand.<>