Visual Hide and Seek
Visual Hide and Seek
复制标题
DOI：
10.1162/isal_a_00269
复制
发表时间：
2019-09
期刊：
影响因子：
--
通讯作者：
Boyuan Chen-;Shuran Song;H. Lipson;Carl Vondrick
中科院分区：
文献类型：
--
作者：
Boyuan Chen-;Shuran Song;H. Lipson;Carl Vondrick
我们训练具身代理玩视觉捉迷藏，猎物必须在模拟环境中导航，以避免被捕食者捕获。我们在环境中放置了各种各样的障碍物，让猎物躲在后面，我们只给代理人使用自我中心的角度对他们的环境进行部分观察。虽然我们从头开始训练模型来玩这个游戏，但实验和可视化表明，智能体学会了预测自己在环境中的可见性。此外，我们定量分析代理的弱点，如速度较慢，影响学习的政策。我们的研究结果表明，虽然代理弱点使学习问题更具挑战性，但它们也会导致学习到更多有用的功能。我们的项目网站可在：这个http URL ~bchen/visualhideseek/.
We train embodied agents to play Visual Hide and Seek where a prey must navigate in a simulated environment in order to avoid capture from a predator. We place a variety of obstacles in the environment for the prey to hide behind, and we only give the agents partial observations of their environment using an egocentric perspective. Although we train the model to play this game from scratch, experiments and visualizations suggest that the agent learns to predict its own visibility in the environment. Furthermore, we quantitatively analyze how agent weaknesses, such as slower speed, effect the learned policy. Our results suggest that, although agent weaknesses make the learning problem more challenging, they also cause more useful features to be learned. Our project website is available at: this http URL ~bchen/visualhideseek/.