Intro

BoltX is a high-performance library for recognition of rock bolts in 3D data sets.

Using a combination of classical image processing, deep neural networks, and custom algorithms, BoltX is able to use stereographic imaging and LIDAR information to measure the location and pattern of rock bolts in a mine. Along with an appropriate hardware platform to perform imaging, BoltX can dramatically reduce the exposure of staff to potentially unsafe conditions, and help to further automate key processes in the resources sector.

Safety is of paramount concern in the mining industry. Bolts are used to stabilise mineshafts and excavations by anchoring the unstable exposed rock to the deeper rock interior. These bolts are typically placed in prescribed patterns, depending on the type of rock and nature of the excavation.

Naturally, it is vital for miners to have an up-to-date, accurate measure of the placement of these bolts and their state of wear-and-tear. However, it is potentially dangerous for miners to perform this inspection in person, as the safety of the conditions they might be exposed to are unknown. In collaboration with a mining maps provider, we developed novel computer vision algorithms to identify rock bolts using LIDAR and stereoscopic images of a mine. This not only improves safety outcomes for miners by minimising their exposure to potentially unsafe conditions, but also saves time by automating the process of verifying that the bolt placement and condition satisfies safety regulations.