In this context, the goal of this position is to investigate the ways
through which an agent can autonomously build state representations of
sensory data which support the learning of sensorimotor and cognitive
skills. The project will build upon the dataset generation methods
developped in the DREAM project. The proposed methods will be validated
in simulated environments and possibly with experiments on physical
robots like Baxter and PR2.
Applicants should hold a PhD in AI or a related field, have a strong
experience in machine learning (dimensionality reduction, deep learning,
bayesian learning), good programming skills and an interest in robotics.
A significant experience in robotics programming is a plus. Speaking or
understanding French is not required.
To apply, please send a CV, a motivation letter and the contact details
of 2 to 3 references to Stéphane Doncieux ([hidden email])
and Alexandre Coninx ([hidden email]) with [Postdoc
application] in the subject of the mail.