Acoustic Traffic Sign Recognition: A Computationally Efficient Alternative to Image-Based Detection

Abstract

Autonomous vehicles rely heavily on image processing techniques for traffic sign recognition, yet these approaches face significant challenges due to environmental conditions, occlusions, and high computational demands. To address these limitations, this paper introduces a novel acoustic-based approach named Noise Pattern Recognition (NPR) system that utilizes sound-based detection to enhance road awareness for autonomous vehicles. Instead of relying on visual inputs, our approach encodes traffic sign information through specially designed road bumps that generate distinct noise patterns when vehicles pass over them. These acoustic patterns, structured similarly to Morse code, are captured by onboard microphones, processed using signal analysis techniques, and converted into binary sequences that correspond to specific traffic signs. The proposed system consists of three key components: a sound recording module, a signal processing module, and a transmission module that relays detected traffic sign information to the vehicle’s control system. Simulation results show the feasibility of this method by demonstrating its robustness against environmental interference and its ability to operate efficiently with minimal computational resources.