An online system for recognizing handwritten symbols from a user specified alphabet is described. The symbols are written on a digitising tablet. When a symbol is subsequently written the system is required to recognize the symbol irrespective of the scale, orientation and position of the written symbol.

The forces transmitted through any cross-section of a machine member such as a robot arm can be reduced to a single force and torque acting on any chosen point of the cross-section. The force and torque can be measured by a sensor that replaces the cross-section. A special feature of the sensor discussed here is that it is two dimensional, i.e. that it can be manufactured as a thin, planar sandwich. This makes for simplicity, compactness, low cost and ruggedness. In this paper we derive the constraints on design that must be met in order for the output to be unbiased with respect to force and torque directions. We show that feasible planar designs exist.

In many proposed applications, robots are required to recognize a relatively small number of object types, each well defined by shape. In such cases, relatively simple pattern recognition methods may be adequate. This paper presents a number of exceptionally simple shape discrimination algorithms which exploit the particularly covenient data provided by the combination of a rotating camera and analog edge finder proposed by Miller and Boie.