… the models on the fly and/or teach new manipulation skills … -parametric mixture models to
avoid the model selection problem … developed generative models to perform manipulation skills …

… Our work contributes hundreds of 13 simulated assets and tasks for articulated and novel
3D object assets, taking a step 14 towards fully autonomous robotic manipulation skill …

… adaptive and predictive models of sensorimotor … generative models that represent joint
distributions over all relevant variables over time. The temporal latent variables in these models …

… Our work contributes hundreds of simulated assets, tasks and demonstrations, taking a
step towards fully autonomous robotic manipulation skill acquisition in simulation. …

… In recent years, progress in deep generative models has produced methods that learn to ‘…
InfoGAN generative model, in this work we learn to imagine goal-directed object manipulation …

… manifold, or predicting full multi-robot manipulation sequences. To achieve accurate sampling,
we combine learned generative models with local optimization for constraint projection. …

… model: In this work, we focus on the transition model and its applications for robotic
manipulation… an approach to learn a generative transition model for robotic manipulation, for both …

… novel approach using a series of modified Generative Adversarial Networks (GANs). Namely,
… types of robotic manipulators, a MICO robotic manipulator and a Fetch robotic manipulator. …

… robotic domains, our approach can be applied to path planning based on the learned skills.
… We treat the latent space of generative models as a Riemannian manifold. This allows us to …

… 5 1.3 Modules in a robotic manipulation pipeline to which this thesis have made … First, we
will investigate how generative models for grasp synthesis play the role of a mechanism for …