We consider a group of mobile agents operating in a mission space that collaborate to solve a general dynamic optimization problem. The agents seek to maximize the total reward collected and minimize their energy cost. We construct a control policy specifying how the agents move subject to constraints due to the geometry of the mission space, the presence of obstacles, their sensing range, their available energy, and the need to avoid collisions with other agents. The mission space is discretized and modeled as a graph. We apply the distributed actor-critic method from [Pennesi and Paschalidis, 2010] properly modified to benefit from least squares temporal difference learning. We present one concrete application: air vehicles flying below a forest's canopy. We demonstrate that the incorporation of bio-inspired features such as eavesdropping, spatial memory, directional sensing, and grouping into the control policy significantly improves its performance compared to a policy from [Paschalidis and Lin, 2011] that included no such features.

 Fermentation processes as a class of biological processes containing the growth of the biomass (bacteria, yeasts) resulting from the consumption of essential substrate supplies (source of carbon, nitrogen, oxygen, etc.) constitute a very delicate challenge from the control point of view. Nonlinearities and complicated dynamics of the biomass growth followed by the creation of various products (from which especially the variety of antibiotics makes the fermentation processes attractive for the industrial utilization) come hand in hand with the attractivity and going along with high level of uncertainty and difficult online measurement of the process variables turn attempts on optimal control of the fermentation process into a rather complicated task. Gradient method suggests a possible way of handling these issues and combined with fresh control strategies (a constant volume strategy and constant vaporization strategy) proves a significantly better performance on a set of numerical experiments than other methods. Moreover, model structure used in the previous work has been modified so that it corresponds with the new optimization strategies which stand for the main contribution of this paper.

 In this article, we present a new approach to the solution of optimization problems with discrete control variables modeled by binary integer programs (BIP). The solution of BIP is computationally demanding when the number of the BIP variables increase. Two instances that increase the number BIP variables in practical applications are the reduction of the discretization sampling time and the increase of the optimization time period. The proposed approach transforms a single BIP optimization into a linear program (LP) and N feasibility BIPs, with less number of variables. The reduction of the number of variables increases the algorithm speed in providing a solution. The approach permits to solve optimization problems with longer time intervals and with a higher number of control variables, while being computationally tractable. A case study on the electricity cost minimization in a pumping station shows the applicability of the method.

 The paper presents a gradient-based algorithm for initial conditions optimization of nonlinear multivariable systems with boundary and state vectors constraints. The algorithm has a backward-in-time recurrent structure similar to the backpropagation-through-time (BPTT) algorithm, which is mostly used as a learning algorithm for dynamic neural networks. It is shown that dynamic parameter optimization problem can be formulated as the initial conditions optimization problem. Further, it is shown that output parameter identification and output controller design problems can be formulated as dynamic parameter optimization problem. The effectiveness of the proposed algorithm is demonstrated on the problem of output identification and control of a nonlinear two-mass torsional system.

 In this paper, the numerical algorithm based on conjugate gradient method to solve a finite-horizon min-max optimization problem arising in the H_infinity control of nonlinear systems is presented. The feedback control and disturbance variables are formulated as a linear combination of basis functions. The proposed algorithm, which has a backward-in-time structure, directly finds very accurate approximations of these feedbacks. Benchmark examples with analytic solutions are provided to demonstrate the effectiveness of the proposed algorithm.

 Mobile-Ad-Hoc-Networks (MANET) are self-configuring networks of mobile nodes, which communicate through wireless links. One of the main issues in MANETs is the mobility of the network nodes: routing protocols should explicitly consider network changes into the algorithm design. MANETs are particularly suited to guarantee connectivity in disaster relief scenarios, which are often impaired by the absence of network infrastructures. This work proposes a proactive routing protocol, developed via Reinforcement Learning (RL) techniques, to dynamically choose the most stable path, basing on GPS information, among the feasible ones and to consequently increase resiliency to link failures. Simulations show the effectiveness of the proposed protocol, through comparison with the Optimized Link State Routing (OLSR) protocol.