Information Constrained Control For Visual Detection Of Important Areas

In this work, we propose a method for detection of locations with subjective significance in the visual environment using Information Constrained Control (ICC). ICC is a model that takes into consideration not only the goal but also the complexity of the control needed to achieve it, characterized by its deviation from default behavior. This concept resonates with the human visual attention system, which includes an interaction between top-down goal oriented pressure and bottom-up processes of common gaze behavior. We start by providing rationale and intuition for ICC based analysis, then formalize it. Specifically, we formalize a mechanism for estimation of subjective significance. Later, we theoretically compare the ICC to the commonly used STD based mechanism, present the latter as a special case of ICC, suggesting an ICC reward visualization mechanism for STD. Finally, we describe our experiment in a real-world driving environment and present empirical finding to support our claim.