Deep learning models for traffic flow prediction in autonomous vehicles: A review, solutions, and challenges

Abstract

In the last few years, there has been an exponential increase in the usage of the autonomous vehicles across the globe. It is due to an exponential increase in the popularity and usage of the artificial intelligence techniques in various applications. Traffic flow predication is important for autonomous vehicles using which they decide their itinerary and take adaptive decisions (for example, turn let or right, move straight, lane change, stop, or accelerate) with respect to their surrounding objects. From the existing literature, it has been observed that research on autonomous vehicles has shifted from the traditional statistical models to adaptive machine learning techniques. However, existing machine learning models may not be directly applicable in this environment due to non-linear complex relationship between spatial and temporal data collected from the surroundings during the aforementioned adaptive decisions taken by the vehicles. So, with focus on these issues, in this article, we explore various deep learning models for traffic flow prediction in autonomous vehicles and compared these models with respect to their applicability in modern smart transportation systems. Various parameters are chosen to have a relative comparison among different deep learning models. Moreover, challenges and future research directions are also discussed in the article.