lifelong reinforcement learning provides a promising framework for developing
versatile agents that can accumulate knowledge over a lifetime of experience
and rapidly learn new tasks by building upon prior knowledge. However, current
lifelong learning methods exhibit non-vanishing regr