Publications - Miscellaneous

It is widely accepted in the control community that the Computer Controlled Systems (briefly, CCSs), have a lot of advantages based primarily on the high reconfigurability of the controller platform and the ability of making complex yet fast decisions. Such positive features make the CCS a useful platform for multitasking control, i.e. using only one single-processor platform to control several plants. Within this framework, the computer has to share its computational time to solve several critical tasks, each one with its priority. Taking into account schedulability and real time operating systems problems, it is just a step forward to realize that tasks with low priority could be interrupted any time, with unpredictable distributions, and it is just a logical intuition that not all the tasks can be at the highest priority. From a control point of view, each control task should compute a control input to the controlled system to prevent instability or to ensure the performances. This field seems to be at the border between the control community and the computer science community and it has not been widely investigated. Some work has been presented introducing more general scheduling models and methods for control systems, where control design methodology takes the availability of computing resources into account and allows the trade–offs between control performance and computer resources utilization, and also introducing the idea of any time CCS without solutions. In literature, the Control Server is introduced, allowing the separation between scheduling and control design by treating the controllers as scalable real–time components. Jitter and latency is included in the model and cannot be taken over. Furthermore, the interrupt time, the I/O operations and the overrun handling are not taken into account, simply imposing only soft real time control tasks. Useful tools for the analysis of real time control performance are the Jitterbug and the True Time that analyze the control performance once the control tasks are implemented as soft real time tasks.

Technical Report

Consider the controlled system $dx/dt = Ax + \alpha(t)Bu$ where the pair $(A,B)$ is stabilizableand $\alpha(t)$ takes values in $[0,1]$ and is persistently exciting. In particular, when $\alpha(t)$ becomes zero the system dynamics switches to an uncontrollable system. In this paper, we address the following question: is it possible to find a linear time-invariant state-feedback, only depending on $(A,B)$ and the parameters of the persistent excitation, which globally exponentially stabilizes the system? We give a positive answer to this question for two cases: when $A$ is neutrally stable and when the system is the double integrator.

We address the problem of detecting faulty behaviors of robots belonging to a multi-agent system. Our objective is to develop a scalable architecture that can be adopted to realize a completely decentralized intrusion detector monitoring the agents' behavior. We want the solution to be independent from the set of ``rules'' describing the interaction among the agents, and from their dynamics; (non-invasive) mainly based on HW/SW components that are already present on-board of each agent. We focus on systems with decentralized cooperation schemes where cooperation is obtained by sharing a set of ``rules'' by which each agent plans its next ``action'' and where some of the agents may act not according to the rules due to spontaneous failure, tampering, or malicious introduction.