From skylight input to behavioural output: a computational model of the insect polarised light compass
We propose a new hypothesis for how insects process polarised skylight to extract global orientation information that can be used for accurate path integration. Our model solves the problem of solar-antisolar meridian ambiguity by using a biologically constrained sensor array, and includes methods to deal with tilt and time, providing a complete insect celestial compass output. We analyse the performance of the model using a realistic sky simulation and various forms of disturbances, and compare the results to both engineering approaches and biological data.
Insect neuroethology of reinforcement learning
Thesis - Doctor of Philosophy. I inversigated how insects form associative memories in their mushroom bodies and how this impacts their olfactory learning, visual navigation, and time-delayed reinforcements tasks.
Predator Evasion by a Robocrab
Robocrab: data-driven adaptation of the evation behaviour in fiddler crabs
Disertation - Master of Science. We create a semi-supervised structure of neural network, inspired by the physiology of neurons in fiddler crabs, and train it to adapt the evasion behaviour of fiddler crabs on potential predators, solving a complicated visuomotor problem (developed in Python using the Theano/Tensorflow-based ‘keras’ library
FI-STAR will establish early trials in the Health Care domain building on Future Internet (FI) technology leveraging on the outcomes of FI-PPP Phase 1.
Digitally capturing unique skills involved in European Traditional Sports and Games. We develope a system that allows 3D reconstruction of a human, caputed by multiple sensors, and analyse their motion with respect to some ground truth.