|Discovery of the optimality criteria for full body human movements using inverse reinforcement learning (Turkey-Slovenya bilateral project)||We study humans’ full body control mechanisms for learning and adaptation in the face of environmental perturbations. We focus on the trade off between the risk of injury, i.e. fall and energy. The obtained human policies will be also transferred to humanoid robots for robust posture control.|
|Convergent Human and Robot Learning for Effective Robot Skill Generation||We study several questions related to human-in-the-loop robot control setups.
|Computational Modeling of Mirror Neurons||
|Developmental Robotics||We study the infant development from a computational viewpoint and implement those on robotics platforms within the context of object manipulation. Particular emphasis is put on a biologically valid developmental progression
In addition we try to understand the impairments in development, such as ASDs by computational modeling
|Sign-Representation of Boolean Functions||This line of research focuses on the polynomial sign representation of Boolean functions. In particular, We try to understand the several properties of such representations over all possible Boolean functions. This work builds upon the theorem I have proven: It is always possible to represent any n-variable BF with 0.75*2^n terms (Oztop 2006, Oztop 2009). This bound is the best known so far; but it is not tight bound, thus further improvements should be possible|