Windshield stone-chips are a common problem for drivers, potentially causing accidents by impairing visibility and contributing to environmental damage if not treated properly. This thesis investigates whether stone-chips can be detected using microphones mounted inside a car, and how accurately their impact positions can be estimated. To this end, 40 stone-chip audio files were recorded while throwing gravel at a moving car, and over 11 hours of robustness data during ordinary driving conditions were collected. Based on this dataset, a signal processing pipeline for simultaneous detection and localization is proposed. The pipeline consists of an adaptive energy detector based on a Kalman filter (KF), a sound source localization step utilizing a time difference of arrival (TDOA)-based sensor fusion framework, and a geometrical model to convert the source position estimate into a binary decision. The proposed method demonstrates promising performance on the gathered data, achieving an F1 score of 0.861 for detection and root mean squared errors (RMSEs) of 1–4 dm for localization. A sensitivity analysis indicates that decent performance can be expected even under more challenging conditions.