Combining modern control and geostatistical techniques for mining machine guidance in geological world framework

The paper describes with examples how tactile data from the cutting heads of rock-cutting mining machines can be successfully processed to produce meaningful images of the surrounding rock environment. Although useful for telerobotic machine guidance, it is believed that improved methods are needed for full machine autonomy in the future. It is suggested that the key to such autonomy lies in the utilisation of the geostatistical covariance structure of the orebody within the location-predicting algorithm. After showing the value of such models for simulation of orebody environments, a means of incorporating geostatistics in the state-estimation procedure is presented. The case-study examined relates to the auto-guidance of longwall coal-shearers.<>