Abstract

Nowadays with the Fourth Industrial Revolution (or Industry 4.0), the automation of traditional manufacturing and industrial practices required the deployment of mobile robots that are involved to accomplishseveral tasks to assist workers in a modular production line. The robots are equipped with several embedded sensors (radar, camera) to analyse the nearby environment, in order to move safely and avoid obstacles. Despite,afterthat this technology does not provide to the robots a dynamical global view of the scene. Thus, the cohabitation between humans and robotscan lead to dangerous situations. In order to ensure security between robots and workers, security zones must be detected dynamically throughout the infrastructure. For that, we will implement algorithms to analyse the scene using the global point of view of the cameranetwork already deployed in the factory. Video analytics allows to exploit automatically the video streams in real time with the aim to detect anomalies and to raise immediately an alarm. To this end, the algorithms detect and track elementsof interest (such aspeople, robot and new object occupyingthe scene) over the time, and alert the robots of the presence of any obstacles in the surrounding area. Where a human is detected close to the robot, his movements will be monitored. Based on a human behaviour analysis, the system will decide whether a new robot 'pathshould be calculated to reach the docking station or to stop completely toavoid any collision. This chapter presents a brief overview of these modern computer vision approaches: to detect objects of interest in video streams, and to localize them in the 3D environment. The purpose of these video analytics is to feed a “planner” indicating dynamically which areas should be avoided by a robots' fleet operating in the production lines.