Deterministic chaos in embodied agents

The aim of the proposed faculty exchange project is to start a scientific collaboration with Professor Takashi Ikegami’s laboratory at the University of Tokyo.

The scientific goal of the visiting period is to establish and rigorously define a protocol in order to apply nonlinear dynamics and deterministic chaos to experiments conducted on human, animals, and artificial embodied agents.

Event dates: 30 Jun 2014 - 31 Aug 2014

During the past two decades, dynamical system (DS) approach has been successfully applied to artificial agents and robotic systems to qualitatively describe their behaviour and cognition. Although DS has proven to be an extremely powerful framework in support of embodied cognitive systems few attempts has been made to move towards a quantitative stance.

Dynamical measures like Lyapunov exponent or correlation dimension explain a wide range of empirical data proving to be valuable predictors. For example it has been shown that they are correlated with gait stability, epileptic seizures, heart arrhythmia, and many other behavioural and neural data. In a work currently in progress we are attempting to correlate Lyapunov exponent and correlation dimension to the ability of an e-puck robot controlled by an artificial neural network to adapt at modifications occurring in the environment after the end of the training process.

The methodology currently under development will be refined during the visiting period and tested in a robotic scenario taking into account the possibility to design an experiment suitable for human or animal participants. The experiment will address the cognitive problem of time perception and conceptualization, a topic that is still controversial in the existing embodied literature.