Controlling Steering Angle for Cooperative Self-driving Vehicles utilizing CNN and LSTM-based Deep Networks

Abstract

A fundamental challenge in autonomous vehicles is adjusting the steering angle at different road conditions. Recent state-of-the-art solutions addressing this challenge include deep learning techniques as they provide end-to-end solution to predict steering angles directly from the raw input images with higher accuracy. Most of these works ignore the temporal dependencies between the image frames. In this paper, we tackle the problem of utilizing multiple sets of images shared between two autonomous vehicles to improve the accuracy of controlling the steering angle by considering the temporal dependencies between the image frames. This problem has not been studied in the literature widely. We present and study a new deep architecture to predict the steering angle automatically by using Long-Short-Term-Memory (LSTM) in our deep architecture. Our deep architecture is an end-to-end network that utilizes CNN, LSTM and fully connected (FC) layers and it uses both present and futures images (shared by a vehicle ahead via Vehicle-to-Vehicle (V2V) communication) as input to control the steering angle. Our model demonstrates the lowest error when compared to the other existing approaches in the literature.

Multi-Task End-to-End Self-Driving Architecture for CAV Platoons

“…Similar systems have been tested in recent years, with training data either taken from real-world datasets [27][28][29][30][31][32] or generated in simulations [33][34][35][36]. Some of this research explored the benefits of including temporal dependencies between consecutive images by adding recurrent layers to the network.…”

“…Application of end-to-end self-driving to vehicle platooning: The above works did not address end-to-end deep learning algorithms that are specifically designed for the application in vehicle platoons. This topic was first addressed in [32,34].…”

“…In 2019, Reference [32] was the first to mention the additional implementation of V2V communication to enhance the performance of end-to-end self-driving. The authors proposed a network architecture that processed images recorded by two cooperative selfdriving vehicles simultaneously.…”

“…With our work, we want to address the limitations of [32,34]. Our contributions are the implementation of an end-to-end self-driving model that:…”

Representing Realistic Human Driver Behaviors using a Finite Size Gaussian Process Kernel Bank

“…The scalability issue, which is due to the high percentage of packet drops in over-occupied and dense traffic situations, has been approached from different perspectives by researchers in vehicular academic and industrial community [1], [2], [3], [4]. From a physical-layer perspective, different modulation and coding schemes (MCS) or specific power control methods have been introduced to address the problem [1].…”