Colorado School of Mines’ Edgar Experimental Mine provides a proving ground for these research activities. Active excavation areas will be used as testing areas to assess perception capabilities in a representative environment.
Mechanical Engineering Assistant Professor Andrew Petruska and Mining Engineering Professor Jamal Rostami were awarded a 3-year project by the National Institute for Occupational Safety and Health (NIOSH) to improve mine safety and miner health by advancing the state of the art of roof bolting automation.
In underground mines, roof bolting is essential to support excavated areas and prevent cave-ins. Roof bolting machines are commonly used to drill boreholes and insert anchor bolts—tasks that used to lead to musculoskeletal injuries in miners. While miners no longer have to do these high-risk activities, they are still needed in the operation to position the machines and change out parts, which exposes them to harmful airborne dust known as respirable crystalline silica (RCS).
“Ironically,” noted Petruska, “the very airborne dust that puts miners’ health at risk is the reason they are required to be near the bolting operation in the first place.” Dust in underground mines causes spectral reflections and high-dynamic-range lighting conditions that prevent the use of traditional camera imaging and machine perception. Only a human can see well enough in this environment to operate the machines.
This is where Petruska and Rostami’s project comes in. Teaming with mining industry leader JOY Global (Komatsu), Petruska and Rostami aim to solve the perception problem and get closer to full roof bolting automation. They will couple newly developed event-based active imaging technologies with dust-penetrating millimeter radar imaging to enable robust 3D perception for the mining environment. The hardware solution will be enhanced with machine learning techniques to create a representation of the roof with the support strap/mesh segmented from the native rock.
“Once we understand the geometry,” Petruska said, “we can employ proven robotic planning and control algorithms to automatically place the bolter for clean drilling and insertion.”
While the research team works on the perception problem, undergraduate Capstone Design teams will explore solutions to automate tasks like changing the bit and drilling rod and placing and securing the bolt.
These combined solutions would greatly reduce the health and safety risks to miners. No longer required for perception and tool changeouts, miners could control the roof bolting equipment remotely, reducing their exposure to hazardous RCS and lowering their risk of being injured by roof falls at the working face.
“The potential outcomes from this project are really exciting,” said Petruska. “Our solution to automate the perception aspect of the roof bolting operation will get the industry a big step closer to full process automation—and, more importantly, it will improve miner health and safety.”