Insights and Strategies for an Autonomous Vehicle With a Sensor Fusion Innovation: A Fictional Outlook

Abstract

A few decades ago, the idea of a car driving without human assistance was something inconceivable. With the advent of deep learning-based machine learning in artificial intelligence, this imaginary idea has become part of our life. Like in other fields, these technological revolutions have brought drastic changes to the field of automated driving systems. The autonomous vehicle is in the transition state between level 3 and level 4 of automation, but many mysteries are still waiting to be solved. Understanding the environment as precisely as a human driver is still far in the future. To attain human perception requires the capturing of extensive surrounding information that depends on the onboard sensors installed on the vehicle. Because the recent autonomous vehicle is equipped with several sensors, it captures surrounding information in diverse forms. Combining these multi-domain data with sensor fusion is the open area of research that is considered in this paper. Along with sensor fusion, another area of prime importance that is necessary to be explored is the prediction of pedestrian intentions. Though the study of the prediction of a pedestrian's intentions started approximately fifteen years ago, most of the research is based on detection rather than intention. Furthermore, this paper also discusses related research in the field of prediction of the pedestrian's intentions. At the end of the article, this review paper includes open questions, challenges, and proposed solutions.

Keywords

• Advanced driver assistance system
• deep learning
• pedestrian intention prediction
• sensor
• sensor fusion