A Comparative Drivability Analysis for Autonomous Robots in underground Mines Using the Entropy and SRM Models

摘要

The mining industry is constantly faced with the dual needs for safety and improved productivity. It is widely recognized that robots can play a significant role in predisaster (pre-emption) and post-disaster (recovery) mine rescue operations. This would inevitably enhance productivity and greatly reduce human exposure to dangerous underground mine environment. Nonetheless, the success of a robot in a mine depends greatly on its visual capability to correctly interpret its immediate environment for navigational purposes. This work serves to assist robots' drivability in an underground mine. A probabilistic approach based on the local entropy is employed. The entropy is measured within a fixed window on a stream of mine frames to compute features used in the segmentation process. We then compare results using the statistical region merging (SRM) approach and evaluate the performance to provide useful qualitative and quantitative conclusions. Different regions of the mine, such as the shaft, stope and gallery, are investigated and results show that a good drivable region can be detected in an underground mine environment.