Computer vision-assisted human-in-the-loop measurements: augmenting qualitative by increasing quantitative analytics for CI situational awareness

Abstract

Many infrastructure problems are reported by the public, yet this can result in human-in-the loop, qualitative measurements and lead to slow response times as quantitative data is needed. Cameras already exist in many settings such as smartphones, or moving objects such as UAV-mounted cameras. Since many critical infrastructure (CI) problems often are first noticed by the general public and then reported, their qualitative descriptions can then be accompanied with quantitative measurements by using the indirect measurement of parameters provided using machine vision. In this paper, the authors propose a framework using Agile IoT to add new modalities to already existing sensors (cameras) such as smartphone devices to determine additional parameters using machine vision. This can result in an increase in situational awareness, and meanwhile, response and repair times can decrease, then the overall infrastructure resilience increases. This has the potential to improve preventative maintenance and increase resilience by increasing situational awareness, so resources can be deployed quickly and efficiently where they are needed. This proposed framework can apply to multiple small infrastructure such as lighting standards, playground structures, signage, access gates and fences, electrical wires, and utility poles and its affixed hardware components. The paper shows a proof-of-concept application of this methodology to the concept of tilt detection, with lean determined from simulated and field images. Quick follow-up to problems at appropriate locations can increase system resilience by quickly enabling solving the problem.