The physical models in this manuscript are typically derived from fundamental physical laws that are established based on observations of real-world phenomena. This section begins by introducing well-known datasets utilized for the development of microscopic traffic models. These datasets encompass both human-driven vehicles and autonomous vehicles, capturing the trajectories of vehicles within a specified investigation zone. It is observed that both human-driven vehicles and autonomous vehicles exhibit trajectories of a two-dimensional nature. However, these trajectories are influenced by the drivers’ maneuvers, which are not explicitly recorded in the datasets. To address this, a method for estimating drivers’ maneuvers, including steering and acceleration/deceleration, is proposed. Furthermore, given that the datasets are collected from diverse locations and countries, an analysis is conducted to explore the variability among the datasets using a predictability index. The results indicate significant behavioral differences among drivers from distinct countries.
Real World Observations, Maneuver Estimation and Behavioral Predictability
Lecture Notes in Intelligent Transportation and Infrastructure
2024-07-27
35 pages
Article/Chapter (Book)
Electronic Resource
English
Satellite maneuver detection and estimation with optical survey observations
Springer Verlag | 2022
|FROM SIMULATION TO REAL WORLD MANEUVER EXECUTION USING DEEP REINFORCEMENT LEARNING
British Library Conference Proceedings | 2020
|REAL- WORLD MANEUVER EXTRACTION FOR AUTONOMOUS VEHICLE VALIDATION: A COMPARATIVE STUDY
British Library Conference Proceedings | 2019
|