文献类型：会议论文

英文题名：Survey of Pedestrian Action Recognition in Unmanned-Driving

作者：Chen, Li[2]; Ma, Nan[1]; Wang, Pengfei[3]; Pang, Guilin[1]; Shi, Xiaojun[1]

通讯作者：Ma, Nan

机构：[1] College of Robotics, Beijing Union University, Beijing, China; [2] Beijing Key Laboratory of Information Service Engineering, Beijing Union University, Beijing, China; [3] Communication and Information Centre, State Administration of Work Safety, Beijing, China

第一机构：北京联合大学机器人学院

会议论文集：Cognitive Systems and Signal Processing - 4th International Conference, ICCSIP 2018, Revised Selected Papers

会议日期：November 29, 2018 - December 1, 2018

会议地点：Beijing, China

语种：英文

外文关键词：Autonomous vehicles - Cognitive systems - Convolutional neural networks - Motion estimation - Pedestrian safety - Surveys

摘要：With the development of unmanned-driving car technology, there are higher requirements for the intelligence, safety and stability of intelligent vehicle driving. Especially in a complex and uncertain environment, the driverless car can accurately detect the pedestrian action, which can effectively realize the autonomous driving of the vehicle. This requires that vehicles detect pedestrians firstly, then identify pedestrian body language and try to understand their intentions, predict pedestrian's actions, which form a good interaction cognition between human and vehicle. In this paper, we give a detailed survey about the recent and state-of-the-art research methods in the filed of human action recognition and discuss their advantages and limitations. We analysis the main framework of motion recognition, and summarize the common datasets of this filed. Finally, suggestions for future research directions are offered, which is expected to benefit the follow research. 2019, Springer Nature Singapore Pte Ltd.