The Power of the Senses:
Generalizable Manipulation from Vision and Touch
through Masked Multimodal Learning

The following are visualizations of the policies trained in the paper using Masked Multimodal Learning (M3L).

Tactile Insertion Environment

Training Enviroment

Generalization Enviroment

Door Opening Environment

Training Enviroment

Generalization Enviroment (10x higher damping and friction)

In-hand Cube Rotation Environment

Training Enviroment

Generalization Enviroment (2x cube mass)