Vision-based Automated Flagging System in Construction

Flaggers are a high-risk profession. They are always required to work closely with the open traffic lanes. Any distracted, speeding or intoxicated drivers might hit them, leading to their injuries and fatalities. From 1980 to 1992, a total of 54 fatalities involving flaggers in the construction industry have been reported. To protect flaggers and reduce their exposure to potential vehicular traffic, previous studies proposed and implemented the Automated Flagger Assistance Devices (AFADs). However, the AFADs have not been widely used in practice due to their costs. In addition to hiring a flagger to remotely operate an AFAD, the cost of an AFAD system alone ranges from $25,000 to $30,000 without the consideration of device maintenance. Instead of creating an assistance device, this paper proposed an automated flagging system (AFS) that can guide the traffic without the need for a flagger available on the site. The proposed system is composed of two modules: information capturing and decision-making. The information module is to monitor traffic conditions and retrieves useful information for the decision-making module to decide which sign (STOP or SLOW) to display in the LED panel. So far, a prototype was developed and tested in a laboratory environment. A vehicle detector was trained and integrated into the prototype. The laboratory test results indicated that the prototype could correctly show the STOP or SLOW sign based on the detection of simulated traffics.