BabyWalk: Going Farther in Vision-and-Language Navigation by Taking Baby Steps
BabyWalk: Going Farther in Vision-and-Language Navigation by Taking Baby Steps
复制标题
DOI：
10.18653/v1/2020.acl-main.229
复制
发表时间：
2020-05
期刊：
影响因子：
--
通讯作者：
Wang Zhu;Hexiang Hu;Jiacheng Chen;Zhiwei Deng-;Vihan Jain;Eugene Ie;Fei Sha
中科院分区：
文献类型：
--
作者：
Wang Zhu;Hexiang Hu;Jiacheng Chen;Zhiwei Deng-;Vihan Jain;Eugene Ie;Fei Sha
学习遵循指令对于视觉和语言导航（VLN）的自主代理至关重要。在本文中，我们研究了代理在从由较短路径组成的语料库中学习时如何导航较长路径。我们表明现有的最先进的代理不能很好地概括。为此，我们提出了 BabyWalk，这是一种新的 VLN 代理，它通过将长指令分解为较短的指令（BabySteps）并按顺序完成它们来学习导航。代理使用特殊设计的内存缓冲区将其过去的经验转化为未来步骤的上下文。学习过程由两个阶段组成。在第一阶段，智能体使用来自演示的模仿学习来完成 BabySteps。在第二阶段，代理使用基于课程的强化学习，通过越来越长的指令来最大化导航任务的奖励。我们创建了两个新的基准数据集（长导航任务），并将它们与现有数据集结合使用来检查 BabyWalk 的泛化能力。实证结果表明，BabyWalk 在多个指标上取得了最先进的结果，特别是能够更好地遵循长指令。代码和数据集发布在我们的项目页面上：https://github.com/Sha-Lab/babywalk。
Learning to follow instructions is of fundamental importance to autonomous agents for vision-and-language navigation (VLN). In this paper, we study how an agent can navigate long paths when learning from a corpus that consists of shorter ones. We show that existing state-of-the-art agents do not generalize well. To this end, we propose BabyWalk, a new VLN agent that is learned to navigate by decomposing long instructions into shorter ones (BabySteps) and completing them sequentially. A special design memory buffer is used by the agent to turn its past experiences into contexts for future steps. The learning process is composed of two phases. In the first phase, the agent uses imitation learning from demonstration to accomplish BabySteps. In the second phase, the agent uses curriculum-based reinforcement learning to maximize rewards on navigation tasks with increasingly longer instructions. We create two new benchmark datasets (of long navigation tasks) and use them in conjunction with existing ones to examine BabyWalk’s generalization ability. Empirical results show that BabyWalk achieves state-of-the-art results on several metrics, in particular, is able to follow long instructions better. The codes and the datasets are released on our project page: https://github.com/Sha-Lab/babywalk.