A paper presented by the researchers this week at the Semantic Technology for Intelligence, Defense and Security conference outside of Washington, D.C., says that the system aims “to approximate human visual intelligence in making effective and consistent detections,” the article says.

The Army-funded research will, the researchers say, be able of “eventually predicting” what’s going to happen next in a scene. The approach relies heavily on advances by machine vision researchers, especially in recognizing stationary and moving objects and their properties, CNET reports.

Oltramari and Lebiere are working to create what they call a “cognitive engine” that can understand the rules by which objects and actions are allowed to interact, the article says. The engine incorporates research – activity forecasting – which tries to understand what humans will do by calculating which physical trajectories are most likely, which “models the effect of the physical environment on the choice of human actions.”