Hangjie Mo1 article listing Hangjie Mo in Nature Communications, indexed by Paperzzz from Crossref and OSF.Reinforcement learning in linear embedding space unlocks generalizable control across soft robot configurations Nature Communications