We address the problem of a set of agents reaching consensus by computing the average of their initial states. We propose two randomized algorithms over a directed communication graph where either a random node broadcast its value or a randomly selected pair of nodes communicate in a distributed fashion. The proposed algorithms guarantee convergence in three important definitions, namely: almost surely, in expectation and in the mean-square sense. We show how the parameters of the algorithm can be optimized to improve the rate of convergence and compare its rates of convergence for directed and undirected graphs.
Autonomous vehicles in GPS-denied areas or when cooperating in missions might have access to bearing and range measurements corrupted by noise, rendering the reachable set to be nonconvex since the measurement set is a segment of an annulus in 2D or a spherical shell in 3D. There are various alternatives that could be used in the literature to over-approximate the set by a convex one. However, given the circular part caused by the range measurement, adopting an exact polytopic description would require an infinite number of hyperplanes. In a similar fashion, using ellipsoids suffers from the same problem due to the hyperplane constraints arising from the bearing part. Moreover, if only bearing measurements are available, the measurement set should be unbounded. Motivated by these observations, we propose a generalization of the definition for constrained zonotopes recently introduced in the literature to also consider the $\ell _{2}$ norm and cones (or any other convex set for that matter) as to represent these sets with less conservatism. Given the exact nature of the propagation, these can serve as a worst-case bounds for the true state which is relevant in some applications such as collision avoidance. In simulations, we also illustrate the performance of the computations to be suitable for relatively small sampling times.
13 April 2023 in IEEE Transactions on Automatic Control
We tackle the problem of a set of agents achieving resilient consensus in the presence of attacked agents. We present a discrete-time reputation-based consensus algorithm for synchronous and asynchronous networks by developing a local strategy where, at each time, each agent assigns a reputation (between zero and one) to each neighbor. The reputation is then used to weigh the neighbors' values in the update of its state. Under mild assumptions, we show that: (i) the proposed method converges exponentially to the consensus of the regular agents; (ii) if a regular agent identifies a neighbor as an attacked node, then it is indeed an attacked node; (iii) if the consensus value of the normal nodes differs from that of any of the attacked nodes' values, then the reputation that a regular agent assigns to the attacked neighbors goes to zero. Further, we extend our method to achieve resilience in scenarios where there are noisy nodes, dynamic networks, and stochastic node selection. Finally, we illustrate our algorithm with several examples, delineating some attacks that can be dealt with by the current proposal but not by the state-of-the-art approaches.