Opportunistic Localization and Reactive Navigation in Confined Environments: Development of automated driving system (ADS) for underground mines

Abstract

As there is no global navigation satellite system (GNSS) coverage underground, it is of great value to the mining industry to design an automated driving system (ADS) capable of navigation and localization without relying on GNSS, infrastructure or prior sensor mapping of the entire environment. An ADS was developed to navigate such confined environments and was verified to be highly accurate in a confined office environment. The proposed ADS utilizes a topological map of the environment represented as a bidirectional graph with nodes and edges corresponding to intersections and corridors respectively. The ADS navigates reactively when travelling along corridors by identifying the road edges and staying within the bounds of the road. When an intersection is detected, the ADS makes a turn or continues along the path depending on its global goal in the topological map. The proposed ADS was also tested on an unconfined environment to evaluate its ability to generalize. While the proposed ADS was unable to navigate in an unconfined environment, it is likely that an ADS designed for confined environments would be able to generalize to unconfined environments if the environment can be transformed into a confined representation.