Mining Contract: Machine Learning Enhanced Perception for Automated or Remote Roof Bolting Operations in Underground Mining

Keywords: Automation

Contract #	75D30121C12206
Start Date	9/1/2021
Research Concept	In addition to mining applications, many confined spaces, underground infrastructures, and disaster areas share similar challenges in relation to machine perception in the mining environment—namely significant levels of airborne particles that inhibit imaging and poor lighting conditions. A technology is needed that will enable augmented and virtual reality for remote operation of mining equipment in such environments and provide a path for a mine of the future in which only machines go underground. Such perception advancements will complement current technologies for positioning and operating underground equipment in a variety of environments, including hard rock, industrial minerals, and coal mining operations.

Contract Status & Impact

This contract is ongoing. For more information on this contract, send a request to mining@cdc.gov.

The primary objective of this contract research is to enable roof bolting automation through the following adavancements:

Machine perception of the geometry and support materials by combining newly available commercial imaging and ranging technologies.
Segmentation of the native rock from support materials, such as wire mesh or steel straps, and identification of acceptable drilling locations using machine learning techniques.
Automated planning of rig placement and drilling operations with proven robotics-motion-planning algorithms given a pre-engineered desired drilling pattern.
Development of the design concepts for automated mechanisms that can handle drilling rods, bolts, plates and nuts, and resin or grouting.

In support of these objectives, researchers will develop a solution that can display an image of the support area that is classified as rock/metal (color-coded) to a remote operator. Then, based on user selection, the angle and position of the drill rig for support placement will be planned, enabling visual monitoring of the drilling operation.

A capstone design goal for this research is to develop mechanisms and automation approaches for drill and bolt handling and changeout to enable fully remote operations and future autonomous platforms. A "stretch goal" is to integrate the solution with a Komatsu-provided automated drilling platform to demonstrate remote or autonomous end-to-end capabilities.

Once demonstrated for the mining environment, it is anticipated that this technology approach and developed algorithms will be adopted by other safety-critical areas.