A Lower Bound on Observability for Target Tracking with Range Sensors and its Application to Sensor Assignment

We present a novel lower bound on the inverse of the condition number of the observability matrix for a system composed on one mobile target sensed by multiple stationary sensors. The actual observability matrix cannot be computed since the control input for the target is not known to the sensors. Instead, we present a lower bound on the observability matrix and its condition number assuming bounded velocity model for the target. We then employ this lower bound for assigning pairs of sensors to track a mobile target. We show empirically that optimizing the lower bound performs comparably with optimizing the (unknown) true condition number.