A kalman filter approach to sensor-based robot control

Abstract

A means of controlling a sensor-based industrial robot using the C programming language is presented. A special purpose C function is described that allows robot move instructions to be qualified by desired sensor readings and uses the total applied correction to update the estimate of pre-taught locations using a Kalman filter. The system allows rapid specification of sensory feedback tasks and is demonstrated on a current industrial research problem; namely the lay-up of carbon-fiber profiles for assembly into satellite antenna dishes.