Estimating Vehicle Delay at Unsignalized Intersections with Gene-Expression Programming

The performance of an unsignalized intersection is typically evaluated based on the delay experienced by vehicles in low-priority movements. Several studies in the literature have examined delays at unsignalized intersections under homogeneous and lane-disciplined traffic conditions. However, it is challenging to find a standard delay model for mixed traffic conditions. To address this gap, the present study proposes mathematical models to estimate average traffic delays for the two least-priority movements: through and right-turning movements (for left-hand drive conditions) from minor streets. The required datasets for this investigation were obtained from thirty unsignalized intersections, encompassing both three-legged and four-legged configurations, with divided and undivided road intersection typologies collected from different parts of India. From data analysis, it was observed that the delay incurred by vehicles in through movements on the minor approach is significantly influenced by the degree of saturation (volume-to-capacity ratio), conflicting traffic volume, crossing pedestrian volume, and intersection crossing distance. Similarly, the delay incurred by vehicles in right-turning movements is significantly influenced by the same list of variables except for the intersection crossing distance. Subsequently, Gene-expression programming (GEP), a novel variant of conventional genetic programming, was successfully utilized to develop delay models for the respective movements. The resultant models demonstrated high prediction accuracies, with coefficient of determination ( R 2 ) values between the predicted and observed delays surpassing 0.93. A comparative analysis corroborated the superior predictive capabilities of the developed models when compared to existing ones.