AF-DROPS: A Drone-Based Fault-Tolerant System for Risk Management in Road Construction Sites

Ensuring worker safety in road construction zones is a critical challenge due to the inherently dynamic and unpredictable nature of these environments. This paper proposes the Adaptive-weighting and Fault-tolerant Drone-based Roadside Operator Protection System (AF-DROPS), a cyber-physical-human system that leverages aerial and ground data collection and elaboration to perform real-time dynamic risk management on road construction sites. The AF-DROPS framework is composed of four communicating subsystems: Aerial Detection (ADS), Ground Detection (GDS), Monitoring and Control (MCS), and Actuation (AS). In the ADS, a drone and a computer vision algorithm are devoted to monitoring the construction site and to measuring the distance of vehicles circulating near the site. A central master unit in the MCS continuously evaluates a dynamic Danger Index whose weighting factors are adapted in real-time. The AS receives control signals from the MCS and activates multimodal alerts (visual, auditory, and tactile) depending on the danger level. In addition, a manual distress signal function is included that ensures fast and reliable communication of workers' alerts. AF-DROPS allows improving workers' safety in road construction sites by the continuous implementation of the main risk management phases, i.e., Risk Analysis, Assessment, Treatment and Communication, and Monitoring and Review. In addition, fault-tolerant control is ensured through system redundancy: if the ADS fails, a smart cone in the GDS seamlessly ensures uninterrupted hazard monitoring and worker protection. Experimental tests show the AF-DROPS effectiveness and performance in realistic environment.