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
27.07.2024
35 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
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
|