Arturo's work is the first result of the ASys technology for online reconfiguration in the robotic control testbed.

More details in the following links:

Dissertation
Presentation

Abstract
In these modern times, we are surrounded by many technical systems that are successfully integrated in our society. However, they are still not able to deal with complex, dynamic and uncertain environments without human intervention. Most of these control systems have not reached yet the required levels of reliability and robustness. A good example is represented by the robotic industry. The majority of robots that actually do some useful work for us consist of industrial robots that perform operations in strictly controlled environments and very well defined missions. However, outside the labs we have not seen many other robots doing anything autonomously but wander. This research work concerns with this topic: augmenting systems resilience using a general approach.
In this work, a robot control system is developed for an autonomous mobile robot that executes a simple but challenging indoor surveillance mission in which device failures may occur. After the full definition of the mission, the control architecture is designed, which consists in the main contribution of this research work. This architecture was inspired by the general cognitive architecture The Operative Mind, and includes a metacontrol layer. The meta-control layer is intended to monitor and reconfigure the system if necessary. This approach may appear very similar to the ones applied to fault-tolerant systems. However, the meta-control layer is defined here with absolute generality, and can be easily augmented to further improve the robustness and the autonomy of the control system, by detecting system's events that are not necessarily related with faults. The control system was implemented and experiments were done to verify that the robot can accomplish its mission completely thanks to the self-reconfiguration capabilities of the system brought by the meta-control layer. Other experiments revealed that this meta-control layer can be easily upgraded to enhance the efficiency of the system.