Coordination is a core problem in multi-robot systems, since it is key to ensuring safety and efficiency. Both centralized and decentralized solutions have been proposed, however, most assume perfect communication. This article proposes a centralized method which removes this assumption, and is suitable for fleets of robots driven by generic second order dynamics. We formally prove that (i) safety is guaranteed if communication errors are limited to delays, and (ii) the probability of unsafety is bounded by a function of the channel model in networks with packet loss. The approach exploits knowledge of the network's non-idealities to ensure the best possible performance of the fleet. The method is validated via several experiments with simulated robots.

This paper describes the implementation in silicon of a retina-like sensor characterized by a space-variant resolution similar to that of the human retina. Besides the actual implementation using CMOS technology, the design peculiarities of retina-like sensors are compared with the use of software or hardware electronic remappers. The paper also describes the realization of a retina-like digital camera and its applications for realtime control and image transmission.

No abstract