Simulation Enabling Autonomous Vehicles to Better Hear Emergency Vehicles 2022-01-0940

Autonomous vehicles need to be able to detect and react to the approach of emergency vehicles, such as fire trucks and ambulances. Exterior acoustic sensors may be deployed to “listen” for sirens of these emergency vehicles. Unfortunately, wind noise from turbulence at cruising conditions can greatly interfere with these exterior sensors. Early assessment of wind noise at alternative sensor locations enables engineers to make design changes to reduce that wind noise, either by choosing quieter sensor locations or by changing the source of interfering turbulence. A computational approach to evaluate the wind noise due to exterior shape at acoustic sensor locations is demonstrated in this paper. By comparing simulated spectra of wind noise at each proposed sensor location to that from a nearby siren signal, designers can rank sensor locations for acoustic detection in different frequency bands. Transient, compressible CFD using the Lattice-Boltzmann Method is applied to an autonomous vehicle model to predict wind noise and siren noise at sensor locations. Detailed flow analysis is performed to identify design changes that reduce the interfering wind noise, and to explain in a visual and intuitive manner why some sensor locations perform better than others.