2. Contents
Introduction
Application of AI in Microscopic Models
Advantages of AI in Microscopic Models
Disadvantages of AI in Microscopic Models
Current State of Research
Future Research
Conclusion
Recommendations
References
3. Introduction
•Artificial Intelligence (AI) As a branch of science and technology, creates intelligent
machines and computer programs to perform various tasks which requires human
intelligence.(Wikipedia) . Artificial Intelligence are now taking over human activities
and are able to study and make human decisions in different conditions. According to
Korb and Nicholson [6], artificial intelligence is the “intelligence developed by
humans, implemented as an artefact”. AI has a very impactive role to play in
microscopic models.
•Car-following models describe the processes in which drivers follow each other in the
traffic stream. The car-following process is one of the main processes in all
microscopic models as well as in modern traffic flow theory.
. According to A khan 2018
Since human drivers are to be replaced by machine for achieving safety, efficiency, and
other objectives,the automation functions are to be designed to replace and potentially
extend human driver capability in processing large volume of data on driving
environment and pattern recognition for decision-making. When fully developed, these
features of the cognitive vehicle can enhance reliability of perception of the driving
environment and quick action in support of safety and other objectives in driving.
4. In 1994, Wewerinke [9] modeled the car-following behavior using the Neural
Networks which are a specific type of machine learning models and concluded
that the proposed model had high performance.
The machine learning-based car-following models attempt to learn the car-
following maneuver of human drivers from a large number of human driver
car-following data. From the given data, the machine learning models can
extract the rules or patterns of drivers’ car-following behavior and capture the
potential relationships among the various variables impacting on car-following
behavior. The studies have proved that the machine learning-based car-
following models have high accuracy in mimicking the car-following
maneuver of human drivers.
The study by(Yang et al., 2019) have also proved that the machine learning-
based car-following models have high accuracy in mimicking the car-
following maneuver of human drivers.
Application of AI in
Microscopic Model
5. •AI improves traffic light system (Vogel et al., 2018)
•AI Easily identifies trends and patterns. Machine Learning
can review large volumes of data and discover specific
trends.(A khan., 2018)
•. AI methodologies are used in predicting traffic congestion by
estimating parameters related to traffic congestion (Akthar &
Moridpour,2021).
Advantages of AI In
Microscopic Model
6. Disadvantages of AI in
Microscopic Model
•The parameters in the model have no physical meanings,
and the outputs of the models are difficult to be controlled in
practice.
•. The machine learning-based car-following models try to
replicate the car-following behavior of human drivers, so the
automated vehicles not only learn drivers’ skill in car-
following but also learn drivers’ inappropriate maneuvers or
even dangerous maneuvers.(Yang et al., 2019)
•
7. Current State Of Research
The machine learning-based car-following models are widely
adopted to control the longitudinal movements of automated
vehicles, such as Google Car and Apple Car, by mimicking
the human drivers’ car-following maneuver. However, like
human drivers, the models easily produce unsafe maneuvers
for automated vehicles and has low robustness, especially in
uncommon situations. There is need for more research to
improve the AI performance
8. Future Research
To improve the machine learning-based car-following models,
YANG etl.:proposes to combine the machine learning models
with the kinematics-based car-following models that can
overcome the shortcomings of machine learning models, using
an optimal combination prediction method, which is called the
combination car-following model. The kinematics-based car-
following model Gipps model has an intrinsic crash-avoidance
mechanism.
9. Conclusion
AI performances in car following models has being a great
achievement in modeling and would be more essential in the
future with respect to technology advancement. By combining
with a collision-avoidance kinematicsbased car-following
model (the Gipps model), the safety level and robustness of
the machine learning-based car-following models are
enhanced for the control of automated vehicles If one model in
the combination model fails in practice, the automated
vehicles controlled by the model still can perform well.
10. REFERENCE
Vogel, A., Oremović, I., Šimić, R., & Ivanjko, E. (2018). Improving Traffic Light
Control by Means of Fuzzy Logic. 2018 International Symposium ELMAR, 51–56.
https://doi.org/10.23919/ELMAR.2018.8534692
Tang, T.-Q., Gui, Y., Zhang, J., & Wang, T. (2020). Car-Following model based on
deep learning and Markov theory. Journal of Transportation Engineering, Part A:
Systems, 146(9), 04020104.
Wu, Y., Tan, H., Chen, X., & Ran, B. (2019). Memory, attention and prediction: A
deep learning architecture for car-following. Transportmetrica B: Transport Dynamics,
7(1), 1553–1571.
Zhou, Y., Fu, R., Wang, C., & Zhang, R. (2020). Modeling Car-Following Behaviors
and Driving Styles with Generative Adversarial Imitation Learning. Sensors (Basel,
Switzerland), 20(18). https://doi.org/10.3390/s20185034