Ideally, robots should be able to learn from very few demonstrations of any given task, and instantly generalize to new situations of the same task, without requiring task-specific engineering.
In this paper, we propose a meta-learning framework for achieving such capability, which we call one-shot imitation learning.
Task examples:
to stack all blocks on a table into a single tower
to place all blocks on a table into two-block towers
A neural net is trained
such that
when it takes as input
the first demonstration demonstration and
a state sampled from the second demonstration,
it should predict the action corresponding to the sampled state.
Our experiments show that the use of soft attention allows the model to generalize to conditions and tasks unseen in the training data