I am interested in understanding (individual and social) cognitive phenomena by building artificial agents capable of solving tasks that required cognitive capabilites. When I say "understanding" cognition, I mean finding operational mechanisms both at the level of agent- environment interaction and at the level of neural structures that guide the development as well as the evolution of cognitive capabilities, including communication and language skills.

To pursue my objectives I generally use:

• Artificial neural networks because:
• they are biologically plausible control structures; therefore, they may tell us something about how biological brains may works;
• with respect to other types of control structures, artificial neural networks require less pre-defined assumptions from the designer, both in terms of structural and functional properties;
• they are robust and capable of generalisation;
• Artificial evolution because:
• it prevents the designer from including incorrect/unnecessary assumptions on how things should be or should work;
• in combination with artificial neural network, artificial evolution represents a reliable method to ground meaning into the sensory-motor experience of the agent;
• Simulation environments because:
• if properly used, it is like working with physical agents. In a sense, I would say that, if simulation is properly done, using a physical agent (robot) does not add anything to what can be found out by using simulation.

A list of potential artificial life/robotic projects can be found below. The projects are fairly flexible and can be adjusted to suit your particular interests. If you have any ideas of your own in these (or related) areas, then please come and see me for a chat!

Compositional semantics in natural language refers to the human ability to understand the meaning of spoken or written sentences from the meaning of their parts, and the way in which these parts are put together. For example, the meaning of an unknown sentence like ``Susan likes tulips'' can be understood by learning the following three sentences: ``Julie likes daisy'', ``Julie likes tulips'', and ``Susan likes daisy''. In this example, the meaning of the original sentence is achieved through compositional semantics by generalising the meaning of single words from a known (already learnt) to an unknown (yet to be learnt) context. The aim of this project is to design artificial agents capable of autonomously develop compositional semantic structures in order to access and correctly execute 'linguistic' instructions.

This project concerns the design of control mechanisms to allow simulated agents to discriminate and categorise different types of objects by interacting with them.

This is about using Evolutionary Robotics techniques to look at the evolutionary circumstances that facilitates the emergence of simple communication protocols among autonomous agents. This research work can be complemented by an analysis of the evolved mechanisms using the tools of dynamical system theory.

This is about using Evolutionary Robotics techniques to look at the circumstances that facilitates the development of learning mechanisms from scratch. It is particularly interesting to look at the selective pressures required to evolved the mechanisms for associative learning. This research work can be complemented by an analysis of the evolved mechanisms using the tools of dynamical system theory.

This is about using spike-neurons to control an autonomous agents required to solve an adaptive task. The challenges concern: 1) the development of methodologies with which to automatically set the parameters of the neural network; 2) the suitability of this type of controllers to guide agents that has to solve tasks that required learning and memory structures.

This project is about designing homogeneous controllers for group of robots required to cooperate and coordinate their actions in order to solve tasks that are beyond the capabilities of a single agent. In particular, I am interested in using range and bearing sensors and communication to develop coordinated actions.

The idea is to develop a system capable of continuously developing the required competencies to face new task/circumstances. Honestly, I have never done any work on this subject. But if you like it, come and have a chat, and we fill find out how to work out a successful project.