Development of electric scooter alerting sounds using psychoacoustical metrics

Abstract

In recent years electric micromobility transportation, including electric scooters, has seen a surge in popularity due to technological advances and the move to lower emission transport. Although offering a range of societal benefits, such as reduced pollution and increased personal mobility, concerns have been raised regarding the implications for pedestrian safety, most notably within the blind and partially sighted community. The issue of pedestrian safety is well studied in the context of larger electric vehicles (EVs), and indeed regulations are now in place that specify mandatory Acoustic Vehicle Alerting Systems (AVAS) for such vehicles. However, limited research has been done on the development of acoustic alerting systems for micromobility. In this paper, the development of an electric scooter (e-scooter) AVAS is considered by taking a perception-influenced design approach to designing alert sounds that optimise detectability and annoyance. A listening experiment has been conducted using ambisonic soundscapes and simulated auralisations of e-scooter passes at 20 km/h, in which a detection-based task and annoyance rating task were conducted. Objective metrics for detectability and annoyance were subsequently evaluated in relation to the subjective responses, so as to enable a more focused approach to the development of alert sounds. Results show that without additional alert sounds, the rate of detection for e-scooters in a soundscape of 60 dBA is as low as 23%. Regression analysis showed that the objective metric of Zwicker’s psychoacoustic annoyance is a useful predictor of subjective annoyance for AVAS sounds, with a coefficient of determination of R2=0.96, and explains more variance than other metrics previously reported in the literature. Partial loudness was also studied as a predictor of detectability, with strong positive association seen (R2≈0.9). Of the alert sounds evaluated, those comprising pure tones with frequency content in the 800 Hz - 1 kHz range, and with amplitude modulation or impulsive characteristics, offered the greatest balance between detectability and annoyance. This study offers much needed research into detectability of electric micromobility transport in a range of environmental noise conditions, and furthermore provides objective metrics for the development of micromobility AVAS sounds going forward.