作者: Zahra Amini , Ramtin Pedarsani , Alexander Skabardonis , Pravin Varaiya
DOI: 10.1109/ITSC.2016.7795752
关键词: Key (cryptography) 、 Bar (music) 、 Stochastic gradient descent 、 Intersection (set theory) 、 Queue 、 Computer science 、 Algorithm 、 Real-time computing
摘要: We consider the problem of estimating queue-lengths at an intersection from a pair advance and stop bar detectors that count vehicles, when these measurements are noisy biased. The key assumption is we know weather queue empty or not. propose real-time estimation algorithm based on stochastic gradient descent. provably learns detector bias, efficiently estimates queue-length with theoretical guarantee. tested in simulation case study using traffic data Beaufort, North Carolina.