Abstract
Humans can master a new task within a few trials by drawing upon skills acquired through prior
experience. To mimic this capability, hierarchical models combining primitive policies learned from
prior tasks have been proposed. However, these methods fail short comparing to the human’s range of
transferability. We propose a method, which leverages the hierarchical structure to train the
combination function and adapt the set of diverse primitive polices alternatively, to efficiently produce
a range of complex behaviors on challenging new tasks. We also design two regularization terms to
improve the diversity and utilization rate of the primitives in the pretraining phase. We demonstrate
that our method outperforms other recent policy transfer methods by combining and adapting these
reusable primitives in tasks with continuous action space. The experiment results further show that our
approach provides a broader transferring range. The ablation study also show the regularization terms
are critical for long range policy transfer. Finally, we show that our method consistently outperforms
other methods when the quality of the primitives varies.
Qualitive Results
Coming soon
Resource and Citation
Paper  
Code  
Video
@inproceedings{aaai21_tseng,
    author = {Wei-Cheng Tseng, Jin-Shang Lin, Yao-Min Feng, Min Sun},
    title = {Toward Robust Long Range Policy Transfer},
    journal = {AAAI},
    year = {2021}
}
Acknowledgement