Urban Wikman, Senior Business Director at Algoryx
“Algoryx was founded in 2007 by four researchers at Umeå University, with the goal of developing a physics engine for VR simulations. Flight simulators have been around for a long time, with physics engines simulating the aerial environment. But it’s much harder to simulate the physics in or under water, in or on the ground, or in space. There are a number of additional factors to consider.
“Our task in the adaptive automation project is to build the actual simulation. We’re using the tools in the Unity game engine to build environments. The simulated drill rig is built from the actual CAD files, with high-precision results based on real data. To make all the physical factors like weights and forces conform to reality, we use AGX Dynamics, which is the physics engine we have developed here at Algoryx. This ensures that the simulated rig and environment behave like they would in real life, and in real time. The values are sometimes known, like the friction between different steel parts or between the wheels and the rock. In other cases you may have to measure on a real rig to get the correct figures.
"The collaboration between the partners is going well. We’ve worked with Epiroc for a number of years now, and they are well aware that they can benefit from the competence of small, agile, high-tech digital companies to further speed up digitalization in the mining industry."
“It’s extremely important that the digital twin is as identical to a real mining setup as possible. If something is off, and you train the system on a faulty simulation, you will have a lot of problems when you start using it in an actual rig.
“At the moment, we are gathering data, scanning mine tunnels and rebuilding them in Unity. The laser scanning process inevitably produces “shadows” in the data, since the laser cannot scan what’s hidden behind something else, and we’re looking at solutions to fill out these empty areas. We’re also creating variations of scenarios, since we need to train the system on a few thousand different scenarios to prepare it for any conceivable problem.
Adaptive automation project
The project aims to enable autonomous face drilling, increasing safety by minimizing human presence at the face. This requires expert AI systems to stand in for operator knowledge. Epiroc leads and coordinates the project and contributes machinery, manpower and expertise. End customer Boliden contributes operations and environment knowledge, testing and demonstration sites, as well as evaluation of solutions. Algoryx provides software development, simulation and Unity expertise, as well as AGX Dynamics, i.e., the physics engine. Örebro University is researching and developing the recognition and machine-learning algorithms.
