1 paper · 1 filter
Yushen Liang, Yue Peng, Baosheng Jin +6
Existing robot policy memories typically evolve with observation sequences, implicitly treating each update as the same amount of physical time. Yet identical observation histories…