Abstract: In this study, we propose a multitask reinforcement learning (RL) algorithm for foundational policy acquisition to generate novel motor skills. Learning the rich representation of the ...