Search

The use of e-scooters and the associated head injuries have increased significantly in recent years. To protect the head and face of the riders, helmets are essential equipment, sometimes mandatory according to the regulations, which makes them an important research target. In this article, we propose a practical helmet monitoring system for e-scooters based on low-cost, little-invasive, real-time estimation of proper helmet wearing while driving. The system is based on both the received signal strength indicator (RSSI) information transmitted by a Bluetooth Low Energy (BLE) tag placed on the helmet and the angle of arrival (AoA) deduced by an antenna array fixed on the scooter. Those measurements feed an efficient software implementation of a statistical estimator specifically designed to deduce at each moment the probability of wearing the helmet correctly. The system has been tested in real indoor and outdoor scenarios by different users to assess its flexibility, robustness, and computational efficiency. We have also performed a comparison with (RSSI) metrical location fingerprinting and ablation analyses. Our approach has shown better performance and coped well with common driving situations (that we also catalog in this work), demonstrating that using statistically filtered signals from AoA plus BLE devices integrated into the same system is an adaptive, reliable, and efficient approach to monitoring safety in e-scooters and possibly other similar vehicles.