Pedestrian Detection System with Edge Computing Integration on Embedded Vehicle

Abstract

This article proposes use of edge computing in detection of pedestrian using a roadside mounted edge computer and conve ying position of pedestrian on road to all nearby vehicles so that vehicles can decide upon possibility of collision of itself with pedestrian and take suitable action to prevent possible accident. This approach of edge computing saves bandwidth, increases reliability and improves response time, and these three parameters are crucial in case of both manual and autonomous driving scenario. Proposed edge computing based system ensures low system cost and higher efficiency as compared to cloud computing architecture. Edge computing is a new computer paradigm that enables real-time safety applications to conduct data analytics at the data source (7, 8). Computer applications, data, and services are moved away from centralised computing infrastructures and closer to users using edge-based computing. For example, by offering computational capability, a roadside data infrastructure positioned in the next immediate edge layer (e.g., roadside traffic infrastructure) from the linked CVs can enable CV safety applications. An edge-centric connected vehicle system, in general, consists of (at least) three edge layers: I mobile edge (e.g., connected vehicles), ii fixed edge (e.g., roadside infrastructures), and iii system edge (e.g., backend server at Traffic Management Center (TMC)).