< Back to previous page
Project
Using invariant theory to extend the extrapolation capabilities of robot learning from demonstration methodologies
This research focuses in the use of virtual demonstration in the robot learning by demonstration field. The objective is to analyse the possible advantages in combining the Coordinate invariant trajectory representation, used to generate virtual demonstrations, with other learning methodologies, as the trajectory parametrised Probabilistic Principal Component Analysis, to increase their generalization capabilities
Date:1 Oct 2021 → Today
Keywords:Robot programming by demonstration, Robot learning by demonstration, Invariant trajectory representation, Virtual demonstrations
Disciplines:Human-centred and life-like robotics
Project type:PhD project