We train a model to learn a Gaussian policy from the input data like images or videos. The Gaussian policy is a policy that modeled by a Gaussian distribution.

For example, the mean of an action (e.g. the torque to a motor and the variance of such distribution).