(234d) Virtual Testing of Image Sensors for Equipment and Operator Safety

The rise of automation has lead to an increase in tasks being handled by machines as opposed to human operators. However, operators perform an essential role in the running of modern plants, performing tasks such as fault detection, supervision of overall functioning of the plant, and being the human-in-the-loop for large automated processes. This has lead to fewer operators that handle some of the most sensitive tasks in a plant’s day to day operation. Image sensors can be a versatile method of maintaining safety, for example, in automotive applications using CMOS image sensors, as in the works of Elkhalili et al. [1] where the sensors are used to determine the distance of objects in the visual field, and in Yamazato et al. [2], which uses the sensors to carry out visual light communication (VLC) to facilitate communication between individual vehicles. The safety of operators in Chemical Engineering has been explored through the modeling of operator behavior [3], and the analysis of operator training [4]. In this paper we demonstrate the use of image sensors as an automated means of maintaining the safety of human operators. These image sensors are simulated virtually using Blender, a free and open source 3D modeling and animation software. A model of a human operator connecting and disconnecting a hose to a tank in a room are created in Blender, and a camera is set up to capture images of the operator, the hose connector, and operators hands as they perform the task. Here we develop three cases of use for image sensors in safety applications: (1) to determine the position of the operator based on images captured by the camera, to function as a location-based safety alert system. (2) to determine the position of the hose connector from the outlet of the tank, to detect if the connector is an unsafe position. (3) to determine if the human operator performs the task in a safe manner by analyzing the haptic information of the operator observed by the camera.

References
[1] Omar Elkhalili et al. “A 64× 8 pixel 3-D CMOS time of flight image sensor for car safety applications”. In: 2006 Proceedings of the 32nd European Solid-State Circuits Conference. IEEE. 2006, pp. 568–571.
[2] Takaya Yamazato et al. “Image-sensor-based visible light communication for automotive applications”. In: IEEE Communications Magazine 52.7 (2014), pp. 88–97.
[3] YM Sebzali and XZ Wang. “Joint analysis of process and operator performance in chemical process operational safety”. In: Journal of Loss Prevention in the Process Industries 15.6 (2002), pp. 555–56
[4] John Haesle, Chris Devlin, and Jack L Mccavit. “Improving process safety by addressing the human element”. In: Process Safety Progress 28.4 (2009), pp. 325–330.