arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1909.12995cs.ROcs.LG
收藏

基于Meta策略优化的快速自适应学习

Learning Fast Adaptation with Meta Strategy Optimization

Georgia Institute of Technology(佐治亚理工学院);Robotics at Google(谷歌机器人部门)

Wenhao Yu, Jie Tan, Yunfei Bai, Erwin Coumans, Sehoon Ha

摘要:

在新场景中行走的能力是足式机器人走向现实世界应用道路上的一个关键里程碑。在这项工作中,我们介绍了Meta Strategy Optimization,这是一种元学习算法,用于训练具有潜在变量输入的策略,可以快速适应目标环境中少量试验的新场景。MSO背后的关键思想是在训练和测试阶段暴露相同的适应过程,即策略优化(SO)。这使得MSO能够有效地学习运动技能以及适合快速适应的潜在空间。我们评估我们的方法在一个真正的四足机器人,并证明成功的适应在各种情况下,包括仿真到现实迁移,步行与削弱电机,或爬上斜坡。此外,我们定量地分析了在模拟环境中的训练策略的泛化能力。真实和模拟实验表明,我们的方法优于以前的方法,在适应新的任务。

英文摘要:

The ability to walk in new scenarios is a key milestone on the path toward real-world applications of legged robots. In this work, we introduce Meta Strategy Optimization, a meta-learning algorithm for training policies with latent variable inputs that can quickly adapt to new scenarios with a handful of trials in the target environment. The key idea behind MSO is to expose the same adaptation process, Strategy Optimization (SO), to both the training and testing phases. This allows MSO to effectively learn locomotion skills as well as a latent space that is suitable for fast adaptation. We evaluate our method on a real quadruped robot and demonstrate successful adaptation in various scenarios, including sim-to-real transfer, walking with a weakened motor, or climbing up a slope. Furthermore, we quantitatively analyze the generalization capability of the trained policy in simulated environments. Both real and simulated experiments show that our method outperforms previous methods in adaptation to novel tasks.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑