Search

Automation of machines in underground mines is a topic with increasing interest, both for research and industrial applications. Autonomous load–haul–dump (LHD) machines need to load material successfully before dumping it into a crusher or an ore pass. The autonomous loading method must be robust to enable reliable operation of the LHD during long periods of time. In this work, a method to perform autonomous loading in Room & Pillar mines is presented. It is based on detecting all ore piles in real‐time, and then computing attack poses in each pile. Then, a positioning process is performed to get the machine in front of the selected ore pile, and an excavation algorithm is executed for loading until the bucket is filled. The proposed method is able to detect multiple ore piles, with different slopes and sizes, and to consider different possible trajectories for attacking and loading the most feasible pile. The method was tested in the Werra Potash Mine, under real operational conditions. The results show that the method can load about 80% of the material that an experienced operator can load. Also, the success rate of the autonomous loading process is very high, being able to load enough material in all trials, and performing successfully the full procedure in 88% of the cases. Thus, the proposed autonomous loading method is a suitable alternative to be used in Room & Pillar mines. [ABSTRACT FROM AUTHOR]

Abstract: This paper addresses the problem of estimating object pose from high‐density LiDAR measurements in unpredictable field robotic environments. Point‐cloud measurements collected in such environments do not lend themselves to providing an initial estimate or systematic segmentation of the point‐cloud. A novel approach is presented that evaluates measurements individually for the evidence they provide to a collection of pose hypotheses. A maximum evidence strategy is constructed that is based in the idea that the most likely pose must be that which is most consistent with the observed LiDAR range measurements. This evidence‐based approach is shown to handle the diversity of range measurements without an initial estimate or segmentation. The method is robust to dust. The approach is demonstrated by two pose estimation problems associated with the automation of a large mining excavator. [ABSTRACT FROM AUTHOR]