Towards Automating Ore Sorting with Rich Sensors

To make mining more efficient and environmentally friendly, transportation of rock should be minimized. Ideally we would only transport those rocks that contain economically extractable amounts of minerals. MineSense has developed sensors that can detect the level of minerals in rocks and, in particular, can detect very low levels with high accuracy. This proposal is to use MineSense’s sensing technology to make decisions about which rocks to keep, and which to discard, as they pass over an array of sensors via a moving belt. Part of the decision is to determine where the rocks are and how many there are. This project will apply some of the latest artificial intelligence techniques, in particular techniques from relational probabilistic modeling, to this problem.

Faculty Supervisor:

David Poole;David Lowther

Student:

Partner:

MineSense Technologies Inc.

Discipline:

Computer science

Sector:

Mining; Professional, scientific and technical services

University:

McGill University; The University of British Columbia

Program: