Iterative Range and Road Parameters Estimation Using Monocular Camera on Highways

Ustunel E., Maşazade E.

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, cilt.24, sa.1, ss.139-151, 2023 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 24 Sayı: 1
  • Basım Tarihi: 2023
  • Doi Numarası: 10.1109/tits.2022.3215892
  • Dergi Adı: IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Communication Abstracts, Compendex, INSPEC, Metadex, Civil Engineering Abstracts
  • Sayfa Sayıları: ss.139-151
  • Anahtar Kelimeler: Roads, Cameras, Three-dimensional displays, Estimation, Iterative methods, Numerical models, Kalman filters, Road parameters estimation, 3D road modeling, monocular camera, least squares, optimization, SENSORS
  • Marmara Üniversitesi Adresli: Evet

Özet

Lane Keeping Assistance and Adaptive Cruise Control in Advanced Driving Assistance Systems (ADAS) requires road parameters for better control of the vehicle. Therefore, in this paper we present the Iterative Least Squares with Optimization (ILSO) method based on a monocular front camera which estimates the horizontal, vertical parameters, lateral slope and the camera’s distances to the left/right line as road parameters and longitudinal range. The ILSO algorithm is smartly divided into two steps LS and optimization depending on their estimation capabilities. First of all, the horizontal and vertical parameters, lateral slope, and distances to the left and right lines are estimated by Least Squares (LS). Then, the longitudinal range is estimated by convex optimization. LS and convex optimization are executed in an alternating manner. ILSO is in advance prior to changes on roads and a simple technique due to its analytical solution which does not require complex geometric calculations and does not need a match between left and right lines. In a simulation environment, given the noisy road lines, ILSO outperforms covariance based and multilayer perceptron (MLP) based methods under different noise variances. In addition, ILSO provides plausible numerical results in both virtual and real road images.