Abstract:
This paper presents a method for single sensor simultaneous derivation of three-dimensional mapping data and material type data for use in an autonomous sandblasting system. A Hokuyo laser range finder's firmware has been modified so that it returns intensity data. A range error and return intensity analyzing algorithm allows the material type of the sensed object to be determined from a set of known materials. Empirical results have demonstrated the system's ability to classify material type (under alignment and orientation constraints) from a set of known materials common to sandblasting environments (wood, concrete, metals with different finishes and cloth/fabric) and to successfully classify objects both when static and when fitted to an in-motion 6-DOF anthropomorphic robotic arm.
