On the Value of Ignorance: Balancing Tracking and Privacy Using a Two-Bit Sensor

We consider a target tracking problem in which, in addition to sensing some information about the position of a mobile target, the tracker must also ensure that the privacy of that target is preserved, even in the presence of adversaries that have complete access to the tracker’s sensor data. This kind of problem is important for many kinds of robot systems that involve communication systems or agents that cannot be fully trusted. In this paper, we (1) introduce a formal, quantitative definition for privacy, (2) describe algorithms that allow a robot to maintain conservative estimates of its performance in terms of tracking and privacy, (3) give strategies for the tracker to maximize its tracking performance, subject to constraints on the allowable privacy levels, and (4) present an implementation of these methods along with some experimental results.