A Survey on Deep Learning for Steering Angle Prediction in Autonomous Vehicles

Abstract

Steering angle prediction is critical in the control of Autonomous Vehicles (AVs) and has attracted the attention of researchers, manufacturers, and insurance companies in the automotive industry. Different Deep Learning (DL) architectures have been applied to predict the steering angle of AVs in various scenarios. A survey on steering angle prediction based on deep learning algorithms can help expert researchers identify those areas that require development. Also, novice researchers can use the survey as a starting point. In this paper, we present a broad study on the recent advances made in DL architectures that covers the steering angle prediction of AVs. A new comprehensive taxonomy of the application of DL in steering angle prediction of AVs is created. The survey presents a concise research summary synthesis, and analysis. It is found that most researchers depend on Convolutional Neural Network (CNN) over other DL architectures in predicting the steering angle of autonomous driving vehicles. Also identified are open research problems. The prominent challenge facing DL-based steering angle prediction of AVs is lack of sufficient real-world datasets, which means that researchers largely depend on data generated from simulated environments. Lastly, alternative viewpoints to solve the identified open research challenges are proposed, pointing towards promising future research directions.

Steering Angle Prediction Techniques for Autonomous Ground Vehicles: A Review

“…Recently, Gidado et al [4] presented a survey paper which covers the following topics: deep learning architectures (deep reinforcement learning and convolutional neural network); application of deep learning architectures for steering angle prediction, longitudinal & lateral control; history and other details of frameworks being used for designing and training ANN architectures; analysis of year-wise and frequency wise publication trend of deep learning applications in steering control. Oussama et al [5] presented a short literature review on computer vision and deep learning approach for steering angle prediction.…”

“…Due to remarkable performance in visual imagery understanding, CNN and its variants are being used for the task of steering angle prediction. Various CNN architectures, differing in number of layers and neurons in each layer, have been explored for the purpose of steering angle prediction [4], [87], [91]- [96].…”

“…The researchers in the domain of ANN-based steering control of AVs heavily rely on artificial datasets for the experimentation. However, the use of artificial datasets has challenges in conducting experiment with the aim of deploying the results in a real-world environment [4]. There is a need for a comprehensive research study covering real world road scenarios with different weather conditions (such as snow, rain, sunny and cloudy), illumination conditions and road types (including unpaved and broken roads).…”

How Many Features is an Image Worth? Multi-Channel CNN for Steering Angle Prediction in Autonomous Vehicles

“…Given the expensive nature of this technology in terms of computational resources and price, it is difficult to assemble into a single package ready for level 4 or 5 deployments. While steering angle predictions have become accurate over time, current models still suffer failures due to high vehicle speeds, sub-optimal obstacle avoidance, and the inability to use RGB input as the sole signalling feature as described in [16]. A solution is presented to the steering angle prediction problem to improve accuracy and robustness and enable a reduction in the amount of sensor hardware, data, and software computation necessary to incorporate into an autonomous vehicle.…”

Optimizing Steering Angle Predictive Convolutional Neural Network for Autonomous Car

“…Once the model learns, it can then autonomously predict steering angles without human intervention. For this type of supervised learning steering angle prediction problem, the convolutional neural network (CNN) and its variants are mostly employed due to their remarkable performance in visual imagery understanding [5]. However, the performance of these networks strongly depends on their architecture, design, and training parameters [6].…”