Detection of Crossing Pedestrians and Control Support in Autonomous Vehicles using Edge-devices

The National Institute of Advanced Industrial Science and Technology aims to implement new means of transportation with autonomous vehicles. In autonomous vehicles, pedestrian recognition and control decisions are one of many important issues. In this study, a machine-learning-based-image analysis method is used to verify control judgment for pedestrian recognition, including their direction and their use of mobile phones. In addition, by using an edge computing system that utilizes multiple edge devices, we are aiming for recognition outside the vehicle that is not affected by communication failures. The results of the demonstration carried out for verification of the same have been reported in this paper.

[1]  Yaser Sheikh,et al.  OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields , 2018, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  Jun Li,et al.  Real-time Pedestrian and Vehicle Detection for Autonomous Driving , 2018, 2018 IEEE Intelligent Vehicles Symposium (IV).

[3]  Wilbert G. Aguilar,et al.  Pedestrian Detection for UAVs Using Cascade Classifiers with Meanshift , 2017, 2017 IEEE 11th International Conference on Semantic Computing (ICSC).

[4]  Mohan M. Trivedi,et al.  Pedestrians and their phones - detecting phone-based activities of pedestrians for autonomous vehicles , 2016, 2016 IEEE 19th International Conference on Intelligent Transportation Systems (ITSC).