Dario Bauso

Non-linear protocols for optimal distributed consensus in networks of dynamic agents

We consider stationary consensus protocols for networks of dynamic agents with fixed topologies. At each time instant, each agent knows only its and its neighbors’ state, but must reach consensus on a group decision value that is function of all the agents’ initial state. We show that the agents can reach consensus if the value of such a function is time-invariant when computed over the agents’ state trajectories. We use this basic result to introduce a non-linear protocol design rule allowing consensus on a quite general set of values. Such a set includes, e.g., any generalized mean of order p of the agents’ initial states. As a second contribution we show that our protocol design is the solution of individual optimizations performed by the agents. This notion suggests a game theoretic interpretation of consensus problems as mechanism design problems. Under this perspective a supervisor entails the agents to reach a consensus by imposing individual objectives. We prove that such objectives can be chosen so that rational agents have a unique optimal protocol, and asymptotically reach consensus on a desired group decision value. We use a Lyapunov approach to prove that the asymptotical consensus can be reached when the communication links between nearby agents define a time-invariant undirected network. Finally we perform a simulation study concerning the vertical alignment maneuver of a team of unmanned air vehicles.

Robust control strategies for multi-inventory systems with average flow constraints

In this paper, we consider multi-inventory systems in the presence of uncertain demand. We assume that (i) demand is unknown but bounded in an assigned compact set and (ii) the control inputs (controlled flows) are subject to assigned constraints. Given a long-term average demand, we select a nominal flow that feeds such a demand. In this context, we are interested in a control strategy that meets at each time all possible current demands and achieves the nominal flow in the average. We provide necessary and sufficient conditions for such a strategy to exist and we characterize the set of achievable flows. Such conditions are based on linear programming and thus they are constructive. In the special case of a static flow (i.e. a system with 0-capacity buffers) we show that the strategy must be affine. The dynamic problem can be solved by a linear-saturated control strategy (inspired by the previous one). We provide numerical analysis and illustrative examples.

Consensus in Noncooperative Dynamic Games: A Multiretailer Inventory Application

We focus on Nash equilibria and Pareto optimal Nash equilibria for a finite horizon noncooperative dynamic game with a special structure of the stage cost. We study the existence of these solutions by proving that the game is a potential game. For the single-stage version of the game, we characterize the aforementioned solutions and derive a consensus protocol that makes the players converge to the unique Pareto optimal Nash equilibrium. Such an equilibrium guarantees the interests of the players and is also social optimal in the set of Nash equilibria. For the multistage version of the game, we present an algorithm that converges to Nash equilibria, unfortunately, not necessarily Pareto optimal. The algorithm returns a sequence of joint decisions, each one obtained from the previous one by an unilateral improvement on the part of a single player. We also specialize the game to a multiretailer inventory system.

Mean field linear quadratic games with set up costs

This paper studies linear quadratic games with set up costs monotonic on the number of active players, namely, players whose action is non-null. Such games arise naturally in joint replenishment inventory systems. Building upon a preliminary analysis of the properties of the best response strategies and Nash equilibria for the given game, the main contribution is the study of the same game under large population. We also analyze the influence of an additional disturbance in the spirit of the literature on H∞ control. Numerical illustrations are provided.

Distributed consensus protocols for coordinating buyers

In this paper, we introduce a distributed consensus protocol for coordinating orders of a network of buyers also called agents/decision makers. Each buyer chooses a different threshold strategy, defining its intention to place an order only if at least other l buyers will do the same. We prove that consensus is reached asymptotically globally and coordination is the same that if the decision making process would be centralized, namely, any decision maker (DM) has access to the thresholds of all other DMs and chooses to order or not. The proposed distributed protocol has the advantage that buyers do not have to communicate their threshold strategy in advance, and consensus is reached without exploring all the possible threshold values.

Robust dynamic cooperative games

Classical cooperative game theory is no longer a suitable tool for those situations where the values of coalitions are not known with certainty. We consider a dynamic context where at each point in time the coalitional values are unknown but bounded by a polyhedron. However, the average value of each coalition in the long run is known with certainty. We design “robust” allocation rules for this context, which are allocation rules that keep the coalition excess bounded while guaranteeing each player a certain average allocation (over time). We also present a joint replenishment application to motivate our model.

Robust Mean Field Games with Application to Production of an Exhaustible Resource

Abstract In this paper, we study mean field games under uncertainty. We consider a population of players with individual states driven by a standard Brownian motion and a disturbance term. The contribution is three-fold: First, we establish a mean field system for such robust games. Second, we apply the methodology to an exhaustible resource production. Third, we show that the dimension of the mean field system can be significantly reduced by considering a functional of the first moment of the mean field process.

Attitude alignment of a team of UAVs under decentralized information structure

In this paper, we discuss nonlinear centralized and decentralized information protocols enabling a team of unmanned air-vehicles (UAVs) to reach consensus regarding the attitude of the formation center. The last is averaged over all UAVs attitudes and not known a-priori. The maneuver consists of an horizontal alignment starting at different attitudes while keeping the formation center constant. During the maneuver, each vehicle controls the vertical rate of climb based on sensed information about the relative attitude of only the nearest vehicles. The rate is bounded by the performance capabilities of the vehicles.

Distributed Consensus in Networks of Dynamic Agents

Stationary and distributed consensus protocols for a network of n dynamic agents under local information is considered. Consensus must be reached on a group decision value returned by a function of the agents’ initial state values. As a main contribution we show that the agents can reach consensus if the value of such a function computed over the agents’ state trajectories is time invariant. We use this basic result to introduce a protocol design rule allowing consensus on a quite general set of values. Such a set includes, e.g., any generalized mean of order p of the agents’ initial states. We demonstrate that the asymptotical consensus is reached via a Lyapunov approach. Finally we perform a simulation study concerning the alignment maneuver of a team of unmanned air vehicles.

Multiple UAV cooperative path planning via neuro-dynamic programming

In this paper, a team of n unmanned air-vehicles (UAVs) in cooperative path planning is given the task of reaching the assigned target while i) avoiding threat zones ii) synchronizing minimum time arrivals on the target, and iii) ensuring arrivals coming from different directions. We highlight three main contributions. First we develop a novel hybrid model and suit it to the problem at hand. Second, we design consensus protocols for the management of information. Third, we synthesize local predictive controllers through a distributed, scalable and suboptimal neuro-dynamic programming (NDP) algorithm.