An extension to Nonlinear Inverse Reinforcement Learning with Gaussian Processes (Levine et al., NIPS 2011) that adds variational inference.

See singletest.m for examples of how to run the algorithm and generate various plots.