Learning to Navigate in Human Environments via Deep Reinforcement Learning
文献类型：会议论文
| 作者 | Xingyuan Gao1,2 ; Shiying Sun1; Xiaoguang Zhao1; Min Tan1
|
| 出版日期 | 2019-12-09 |
| 会议日期 | 2019-12-12至2019-12-15 |
| 会议地点 | Sydney, Australia |
| 英文摘要 | Mobile robots have been widely applied in human populated
environments. To interact with humans, the robots require the capac
ity to navigate safely and effiffifficiently in complex environments. Recent
works have successfully applied reinforcement learning to learn socially
normative navigation behaviors. However, they mostly focus on model
ing human-robot cooperations and neglect complex interactions between
pedestrians. In addition, these methods are implemented using assump
tions of perfect sensing about the states of pedestrians, which makes
the model less robust to the perception uncertainty. This work presents
a novel algorithm to learn an effiffifficient navigation policy that exhibits
socially normative navigation behaviors. We propose to employ convo
lutional social pooling to jointly capture human-robot cooperations and
inter-human interactions in an actor-critic reinforcement learning frame
work. In addition, we propose to focus on partial observability in socially
normative navigation. Our model is capable to learn the representation of
unobservable states with recurrent neural networks and further improves
the stability of the algorithm. Experimental results show that the pro
posed learning algorithm enables robots to learn socially normative navi
gation behaviors and achieves a better performance than state-of-the-art
methods. |
| 会议录出版者 | Springer |
| 语种 | 英语 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/47407]  |
| 专题 | 自动化研究所_复杂系统管理与控制国家重点实验室_先进机器人控制团队 多模态人工智能系统全国重点实验室 |
| 通讯作者 | Shiying Sun |
| 作者单位 | 1.The State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences 2.University of Chinese Academy of Sciences |
| 推荐引用方式 GB/T 7714 | Xingyuan Gao,Shiying Sun,Xiaoguang Zhao,et al. Learning to Navigate in Human Environments via Deep Reinforcement Learning[C]. 见:. Sydney, Australia. 2019-12-12至2019-12-15. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。