Self-Adaptive Driving in Nonstationary Environments through Conjectural Online Lookahead Adaptation
Tao Li,Haozhe Lei,Quanyan Zhu,Tao Li,Haozhe Lei,Quanyan Zhu
Powered by deep representation learning, re-inforcement learning (RL) provides an end-to-end learning framework capable of solving self-driving (SD) tasks without manual designs. However, time-varying nonstationary environments cause proficient but specialized RL policies to fail at execution time. For example, an RL-based SD policy trained under sunny days does not generalize well to rainy weathe...
