Visual Servoed Autonomous Landing of an UAV on a Catamaran in a Marine Environment.

This paper introduces a procedure for autonomous landing of a quadrotor on an unmanned surface vehicle in a marine environment. The relative pose and velocity of the vehicle with respect to the quadrotor are estimated using a combination of data coming from a vision system, which recognizes a set of AprilTags located on the vehicle itself, and an ultrasonic sensor, to achieve further robustness during the final landing phase. The considered software and hardware architecture is provided, and the details about the landing procedure are presented. Software-in-the-loop tests were performed as a validation step for the proposed algorithms; to recreate realistic conditions, the movements of the landing platform have been replicated from data of a test in a real marine environment. In order to provide further proof of the reliability of the vision system, a video sequence from a manual landing of a quadrotor on the surface vehicle in a real marine environment has been processed, and the results are presented.

Moving-horizon state estimation for nonlinear systems using neural networks.

Moving-horizon (MH) state estimation is addressed for nonlinear discrete-time systems affected by bounded noises acting on system and measurement equations by minimizing a sliding-window least-squares cost function. Such a problem is solved by searching for suboptimal solutions for which a certain error is allowed in the minimization of the cost function. Nonlinear parameterized approximating functions such as feedforward neural networks are employed for the purpose of design. Thanks to the offline optimization of the parameters, the resulting MH estimation scheme requires a reduced online computational effort. Simulation results are presented to show the effectiveness of the proposed approach in comparison with other estimation techniques.

Active state estimation for nonlinear systems: a neural approximation approach.

In this paper, we consider the problem of actively providing an estimate of the state of a stochastic dynamic system over a (possibly long) finite time horizon. The active estimation problem (AEP) is formulated as a stochastic optimal control one, in which the minimization of a suitable uncertainty measure is carried out. Toward this end, the use of the Renyi entropy as an information measure is proposed and motivated. A neural control scheme, based on the application of the extended Ritz method (ERIM) and on the use of a Gaussian sum filter (GSF), is then presented. Simulation results show the effectiveness of the proposed approach.

Neural approximation of open-loop feedback rate control in satellite networks.

A resource allocation problem for a satellite network is considered, where variations of fading conditions are added to those of traffic load. Since the capacity of the system is finite and divided in finite discrete portions, the resource allocation problem reveals to be a discrete stochastic programming one, which is typically NP-Hard. In practice, a good approximation of the optimal solution could be obtained through the adoption of a closed-form expression of the performance measure in steady-state conditions. Once we have summarized the drawbacks of such optimization strategy, we address two novel optimization approaches. The first one derives from Gokbayrak and Cassandras and is based on the minimization over the discrete constraint set using an estimate of the gradient, obtained through a "relaxed continuous extension" of the performance measure. The computation of the gradient estimation is based on infinitesimal perturbation analysis (IPA). Neither closed forms of the performance measures, nor additional feedbacks concerning the state of the system and very mild assumptions about the stochastic environment are requested. The second one is the main contribution of the present work, and is based on an open-loop feedback control (OLFC) strategy, aimed at providing optimal reallocation strategies as functions of the state of the network. The optimization approach leads us to a functional optimization problem, and we investigate the adoption of a neural network-based technique, in order to approximate its solution. As is shown in the simulation results, we obtain near-optimal reallocation strategies with a small real time computational effort and avoid the suboptimal transient periods introduced by the IPA gradient descent algorithm.