Optimal UAV coordination for target tracking using dynamic programming

Abstract

Abstract-This work focuses on optimal routing for two camera-equipped UAVs cooperatively tracking a single target moving on the ground. The UAVs are small fixed-wing aircraft cruising at a constant speed and fixed altitude; consequently, the vehicles are modeled as planar Dubins vehicles. A perspective transformation, relating the image-plane measurements to the ground, allows derivation of the geolocation (target localization) error covariance. Using dynamic programming, we compute optimal coordinated control policies which minimize the fused geolocation error covariance. A surprising result, and the main contribution of this work, is that the dominant factor governing the optimal UAV routes is coordination of the distances to the target, not of the viewing directions as is traditionally assumed.

Consistent approximation of a nonlinear optimal control problem with uncertain parameters

“…Such a problem formulation allows a broad range of existing control problems to be extended to incorporate parameter uncertainty. For instance, in a number of optimal control applications such as asset protection (Ding, Rahmani, & Egerstedt, 2009) and target tracking (Quintero, Papi, Klein, & Chisci, 2010), the objective functional depends on other agents whose behavior may involve parameter uncertainty. Another application which can be addressed using this formulation is optimal path planning in uncertain environments, such as aircraft routing in a threat environment (Zabarankin, Uryasev, & Pardalos, 2002) or navigating an unmanned surface vehicle in a riverine environment (Gadre, Du, & Stillwell, 2012).…”

Vision-based target tracking with a small UAV: Optimization-based control strategies

“…The vision-based measurements of the target position not only suffer from errors proportional to the UAV's distance from the target as detailed in [22], but they are also corrupted by outliers. Accordingly, the second author has been developing a robust filter that utilizes 1 regularization to reject outliers and simultaneously provide a filtered estimate of the target state.…”

“…The final venue for future work is to extend our work to two UAVs as in [22], but with the gimbal constraints of Section 2.2 and the stochastic kinematics of Sections 3.2 and 3.3. The goal would be for the two UAVs to coordinate their distances to the target to gather the best joint visionbased measurements, which entails that at least one UAV is always relatively close to the target with the target in its field of regard.…”

“…This means that the UAV wants to be as close to the target as possible (to maximize the resolution of the target in the image plane and reduce the error in the vision-based measurements of the target's position [22]), while simultaneously avoiding extreme gimbal angles. This can be captured by a cost objective for the game to be The camera's instantaneous field of view (FOV) and total field of regard (FOR) are indicated by the dark and light gray regions, respectively.…”

“…While such work typically uses nonlinear feedback control, other have used optimal control for the flight coordination of two UAVs. For example, Stachura et al develop an online receding horizon controller in [21] to maximize the total geolocation information in the presence of communication packet losses, and in [22] the authors determine optimal control policies offline that minimize the joint geolocation error over a long planning horizon.…”