Advances in Deep Learning for Autonomous Vehicle Perception: A Comprehensive Review

Namitha  Kalakunnath; Aneesh Varghese; Abekaesh Prakash Anuradha; Dhanush Kumar Girish; Renjith  Sasidharan

doi:10.47852/bonviewJCCE52025836

Authors

Namitha Kalakunnath Department of Computer Science and Engineering, Amrita Vishwa Vidyapeetham Amritapuri, India https://orcid.org/0000-0003-3801-3117
Aneesh Varghese Amazon Web Services Texas, USA
Abekaesh Prakash Anuradha Department of Computer Science and Engineering, Amrita Vishwa Vidyapeetham Amritapuri, India https://orcid.org/0009-0009-8415-6865
Dhanush Kumar Girish Department of Computer Science and Engineering, Amrita Vishwa Vidyapeetham Amritapuri, India
Renjith Sasidharan Department of Computer Science and Engineering, Amrita Vishwa Vidyapeetham Amritapuri, India https://orcid.org/0009-0005-2803-2703

DOI:

https://doi.org/10.47852/bonviewJCCE52025836

Keywords:

autonomous vehicle, perception, object detection, sensor fusion, object localization, deep learning

Abstract

Autonomous vehicle (AV) perception tasks are critical for enabling self-driving cars to navigate complex environments, relying on advanced technologies to interpret and understand the surrounding world through sensors, deep learning models, and sensor fusion techniques. This review paper provides a comprehensive overview of deep learning architectures applied to AV perception tasks, with a particular focus on recent advancements from 2019 to 2024. The paper begins by examining 3D object detection techniques, exploring the state-of-the-art methods developed during the past six years. Moreover, object localization innovations are discussed, pointing out certain key advancements in that area. The paper also discusses sensor fusion techniques and how they are central to improving performance in 3D object detection. Finally, the discussion encompasses various environmental perception methods such as road and lane detection and traffic sign recognition, all of which are crucial for the safe and efficient operation of AVs. This paper aims to provide insights into the evolving landscape of AV perception and its applications in intelligent transportation systems.

Received: 3 April 2025 | Revised: 9 June 2025 | Accepted: 19 June 2025

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.

Author Contribution Statement

Namitha Kalakunnath: Conceptualization, Validation, Writing - review & editing, Visualization, Supervision, Project administration. Aneesh Varghese: Supervision, Project administration. Abekaesh Prakash Anuradha: Methodology, Investigation, Resources, Data curation, Writing - original draft, Writing - review & editing. Dhanush Kumar Girish: Investigation, Data curation, Writing - original draft. Renjith Sasidharan: Writing - review & editing, Visualization.