Most robot manipulators are poor at measuring and controlling forces, especially force of contact. At the innermost control loops these force estimates are needed to control compliance. At a more general level identifying when contact occurs, and the forces involved, are elemental requirements in building cognitive information that allow robots to adapt to an unstructured environment. This paper shows that reliable internal force sensing can be achieved by using information that is already available in the control loop, and using methods based on system identification to estimate the force at the joint level, and hence by implication the end point force. The method relies on the concept of backdrivable linkage transmissions, a linkage design method that is well established in the field of haptic interface design. It is argued that these internal models should be more widely exploited in robotics and are no different to the internal models built by animals and humans that enable us to adapt rapidly in an unstructured world.
展开▼