Acceleration and deceleration models for two-lane two-way undivided roads using naturalistic driving data

Despite a significant portion of roadways being two-lane roads, the literature on acceleration/deceleration (A/D) behaviour on these roads is sparse, more so in mixed traffic. Two-lane two-way undivided roads are unique due to narrower carriageway widths and interactions with on-coming traffic. A/D behaviour on these roads is affected by various characteristics such as driver, vehicle, and section types. In mixed traffic conditions, due to the high degree of heterogeneity in vehicle types and lane-less traffic movement, analysing A/D behaviour by accounting for these characteristics becomes even more important. In the present study, a large volume of naturalistic driving data (410 drivers) is collected in mixed traffic roads covering varying levels of driver, vehicle and section types through the use of multiple onboard sensors. Analyses include variation in A/D behaviour on two-lane two-way undivided roads across various characteristics. The limitations of the literature are addressed in the present work by considering driver, vehicle, and section variables in the A/D models. A/D models segmented by traffic volume period are found to be superior to the combined (peak and off-peak) A/D models. Results show that over 90% improvement in R2 values is observed for acceleration models considering driver, vehicle, and section variables. It is also interesting to note that, driver and vehicle characteristics influence the A/D behaviour more in the peak periods than the off-peak periods. The developed models are found to be useful in estimating fuel consumption and emission rates. The findings from the study can be applied in developing driver assistance and warning systems. © 2023, Aracne Editrice. All rights reserved.