Over 270,000 fatal pedestrian traffic accidents occur yearly worldwide, with most happening during road crossing. Causes include misjudging crossing gaps and time, low visibility, and noticeability. Electric vehicles (EVs), due to their lack of combustion engine and quiet nature, pose a challenge for vulnerable road users (VRUs). Augmenting EV sound levels may aid safety but add to urban noise pollution. The goal of this project is to explore the design of optimal synthetic sound signals for EVs to provide maximum information and noticeability to VRUs while keeping noise pollution to a minimum. We will focus on the V2X (vehicle-to-everything) future scenario, where EVs detect VRUs and emit auditory signals. To address the general issue of high level of annoyance of auditory feedback in automotive quantitatively (in contrast to the prevailing use of subjective measures in the field), we will use SQAT, an open-source sound qualityanalysis toolbox for MATLAB. This approach enables the evaluation of many audio signals within the design loop, which is unfeasible via subjective assessments, such as psychoacoustic listening experiments. These experiments will be performedonce the number of signals is reduced to a feasible value at the newly developed psychoacoustic listening laboratory of TU Delft. Generative AI and ANSYS will assist in signal design, and user evaluations will be compared with AI-generated responses to explore the need for traditional user studies in the era of generative AI.
- Dr. P. Bazilinskyy, assistant professor, Department of Industrial Design of TU Eindhoven
- Dr. R. Merino-Martinez, assistant professor, Aircraft Noise and Climate Effects Section of TU Delft
- Dr. K. Bennin, assistant professor, Information Technology Group of Wageningen University & Research
- Prof.dr. M. Martens, full professor, Department of Industrial Design of TU Eindhoven
- Prof.dr. B. Tekinerdogan, chair, Information Technology Group of Wageningen University & Research