Nice work. They still use a mid-level controller to translate the output of the policy into kinematics which is an all-too-frequently accepted part of robotics.

Nice work. They still use a mid-level controller to translate the output of the policy into kinematics which is an all-too-frequently accepted part of robotics. Ideally control is learned end-to-end, which is how you get more natural robot motion.