1 paper · 1 filter
Jiaqi Wang, Zhuo Zhang, Haining Guan +15
Modern driving action models are increasingly improved in a self-improvement loop, where a learned world simulator imagines future observations and the resulting data is fed back t…