This paper introduces a method for planning the motions of mobile sensors so they both track a target and respect the sensor’s motion limits at all times. The key idea is to make the path planner “estimation-aware.” That means the planner does not treat sensing as perfect data. Instead it explicitly considers how uncertain the target estimate will be as the sensor moves.

The authors build an optimal-control formulation that includes the nonlinear dynamics of the mobile sensor and of the target. They also include the nonlinear evolution of the sensor’s estimate uncertainty, as described by an extended Kalman filter (EKF). The EKF provides a running covariance, which is a measure of how uncertain the estimate is. The method also allows for nonlinear measurement models, so it can work when the way the sensor observes the target is not a simple linear function.

At a high level, the planner optimizes a sensor trajectory while propagating the EKF covariance forward in time. That covariance then shapes the planner’s objective so the sensor moves in ways that reduce uncertainty where needed. A notable technical point is that the method enforces the sensor’s constraints—such as limits on position or velocity—in continuous time. In practice this means the planner tries to avoid constraint violations not only at discrete planning points but between them as well. The authors report that this holds to within numerical precision.

The paper reports numerical simulations in which the method achieves estimation-aware tracking and respects the continuous-time constraints even when the time discretization is sparse (that is, when the planner evaluates only a few time points). This matters because planning that ignores estimation uncertainty can lead to poor tracking or to constraint violations between planning instants. Ensuring constraints in continuous time makes the planned motions safer and more reliable in simulation.