Simple, Robust Autonomous Grasping in Unstructured Environments

The inherent uncertainty associated with unstructured grasping tasks makes establishing a successful grasp difficult. Traditional approaches to this problem involve hands that are complex, fragile, require elaborate sensor suites, and are difficult to control. In this paper, we demonstrate a novel autonomous grasping system that is both simple and robust. The four-fingered hand is driven by a single actuator, yet can grasp objects spanning a wide range of size, shape, and mass. The hand is constructed using polymer-based shape deposition manufacturing, with joints formed by elastomeric flexures and actuator and sensor components embedded in tough rigid polymers. The hand has superior robustness properties, able to withstand large impacts without damage and capable of grasping objects in the presence of large positioning errors. We present experimental results showing that the hand mounted on a three degree of freedom manipulator arm can reliably grasp 5 cm-scale objects in the presence of positioning error of up to 100% of the object size and 10 cm-scale objects in the presence of positioning error of up to 33% of the object size, while keeping acquisition contact forces low.