1 paper · 1 filter
Renju Feng, Rukang Wang, Ning Xi +4
Traditional end-to-end autonomous driving frameworks frequently suffer from the "style-averaging" dilemma when trained on high-variance human demonstrations, yielding homogenized,…