Automatic detection of traffic incidents using data obtained from sensors embedded in intelligent freeways

This paper reviews innovative research done during the past few years on automatic detection of traffic incidents by the author and his associates using data obtained from sensors embedded in intelligent freeways. A multi-paradigm intelligent system approach is employed to solve the complicated and chaotic pattern recognition problem using neural networks, fuzzy logic, and wavelets. Wavelet-based de-noising and feature extraction techniques are employed to eliminate undesirable fluctuations in observed data from traffic sensors. The result is reliable algorithms with high incident detection and very low false alarm rates.