1 paper · 1 filter
Jilei Mao, Jiarui Guan, Yingjuan Tang +7
The visuomotor policy can easily overfit to its training datasets, such as fixed camera positions and backgrounds. This overfitting makes the policy perform well in the in-distribu…