Additional efficient warning sounds for electric and hybrid vehicles

Electric and hybrid vehicles are very silent at low speeds (below 40 km/h). This may represent a danger for pedestrians, especially vulnerable ones (elderly or visually impaired people). Therefore, warning sounds may be used in order to alert pedestrians that a car is approaching. Recently, some recommendations have been defined by the US NHTSA and the European parliament in order to design such warning sounds. These recommendations may encourage the use of loud warning sounds, so that the benefit of electric vehicles for a quieter environment can no longer be upheld. This work was carried out within the eVADER project and aimed to define efficient warning sounds, i.e. sounds which can be easily detected at a low sound pressure level. To that effect, experiments were conducted, in which different warning sounds were added to a recording of a car passing by (at 20 km/h), embedded in an urban background noise. Sounds were presented to listeners through headphones in the laboratory, in a multi-trial procedure. More than one hundred people participated, including 36 visually-impaired persons. This experiment showed that some warning sounds can make the electric car as easily detectable as a diesel car, while the peak level of its noise is as much as 5 dB(A) lower than that of the conventional vehicle. Sound level alone can not explain this result. Efficient warning sounds exhibited high temporal variations which drew the attention of the listeners. In addition, a small number of frequency components (three) proved to be favourable, as this allowed a higher signal-to-noise ratio for a given overall loudness. This experiment proved that warning signals can be designed which allow a compromise between pedestrian safety and environmental noise quality by specifically including human sound perception in addition to sound level. This may imply, however, that current regulations have to be modified since these are still based on the typical conventional engine signal characteristics (level or timbre).