Unifying ground and air - a comprehensive review of deep learning-enabled CAVs and UAVs

Oct 3, 2026

The paper reviews how advances in artificial intelligence, especially computer vision and image recognition, are reshaping intelligent transportation for smart cities through connected autonomous vehicles (CAVs) and unmanned aerial vehicles (UAVs). It emphasizes real-time situational awareness enabled by data fusion and V2X-linked cooperation between AI-augmented platforms. It also addresses the difficulty of selecting deep learning models for connected vehicles and the serious impact of model errors, plus cybersecurity risks requiring timely threat analysis. The study highlights AI-driven fusion gains, security vulnerabilities, and corner-case situational improvements.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages72

Reading Time3 h 1 min

AccessFree to download

Introduction

Research motivation

Unifying ground and air: a comprehensive review of deep learning-enabled CAVs and UAVs Muhammad Umer Zia1 · Wei Xiang2 · Tao Huang1 · Jameel Ahmad3 · Jawwad Nasar Chattha4 · Ijaz Haider Naqvi5 · Faran Awais Butt6 Received: 24 September 2024 / Accepted: 7 October 2025 © The Author(s) 2025 Abstract The tremendous advancements in artificial intelligence (AI) techniques, particularly those pertinent to computer vision and image recognition, are revolutionizing the automotive industry towards the development of intelligent transportation systems for smart cities. Integrating AI techniques into connected autonomous vehicles (CAVs) and unmanned aerial vehicles (UAVs) and their data fusion, enables a new paradigm that allows for unparalleled real-time awareness of the surrounding environment. The potential of emerging wireless technologies can be fully exploited by establishing communication and cooperation among AI-augmented CAVs and UAVs. However, configuring appropriate deep learning (DL) models for connected vehicles is a complex task. Any errors can result in severe consequences, including loss of vehicles, infrastructure, and human lives. These systems are also susceptible to cyber attacks, necessitating a thorough and timely threat analysis and countermeasures to prevent catastrophic events. Our findings highlight the effectiveness of AI-driven data fusion in enhancing cooperative perception between CAVs and UAVs, identify security vulnerabilities in DL-based systems, and demonstrate how V2X-enabled UAVs can significantly improve situational awareness in corner cases. Keywords Deep learning · Artificial intelligence · Connected and autonomous vehicles · Unmanned aerial vehicles · Cybersecurity 1 Introduction The last decade has seen immense progress in making the dream of connected and autonomous vehicles (CAVs) a reality. Deep learning (DL) is undoubtedly the primary technology behind many breakthroughs in image recognition, and robotics (Bonsignorio et al. 2020) . The success of DL techniques in the mentioned fields has led to widespread deployment of this technology with the aim of passenger safety, elimination of roadside accidents, and optimal path planning in self-driving cars (Grigorescu et al. 2020; Kuutti et al. 2020; Rao and Frtunikj 2018; Ni et al. 2020) . The automotive industry has started testing CAVs on“controlled” roads with different capabilities termed “scales” graded from zero to five. The Extended author information available on the last page of the article 1 3 lower scales feature basic driver assistance, while higher scales indicate a vehicle that needs no human intervention in driving (Yurtsever et al. 2020). A complete CAV system combines technologies, sensors, algorithms, and communication infrastructure. The involvement of DL blocks in a CAV system also depends on its scale. A fully automated car operating on scale five may have a distinct DL module attached to its key decision and control systems. To complement CAVs in challenges like surveillance, acquiring aerial data, and combating emergencies, a promising solution is to adopt unmanned aerial vehicles (UAVs)(Hildmann and Kovacs 2019; Amer et al. 2020; Moukahal et al. 2020; Guillen-Perez and Cano 2018) . A UAV is an unmanned autonomous or semi-autonomous machine that can be controlled remotely and allows us to monitor activities at different locations. UAVs can play a vital role in assisting a CAV’s network in conjunction with Vehicle-to-Everything (V2X) communication technology and other advanced network technologies, such as software-defined networking, network function virtualization, mobile edge computing (MEC), and fog computing (Mishra and Natalizio 2020) . Recently, UAVs’ applications in the communications domain, along with their challenges and open problems, are investigated in Mozaffari et al. (2019) . Due to UAV’s versatile nature, automation, and low cost, it enjoys widespread use in civilian applications like surve

FAQs

What problem does the review aim to address about CAVs and UAVs?

It addresses a literature gap by focusing on an integrated view of deep learning-assisted connected vehicles, combining the complementary capabilities of CAVs and UAVs, especially UAV aerial views for holistic traffic understanding.

Why is configuring deep learning models in connected vehicles challenging?

Configuring suitable DL models is complex, and errors can lead to severe consequences such as vehicle loss, infrastructure damage, and risks to human lives.

How can UAVs improve situational awareness for CAV networks?

By enabling cooperative perception through AI-driven data fusion and, with V2X-enabled UAVs, improving situational awareness in corner cases and supporting roles such as flying or emergency roadside units.

Oct 3, 2026

Upload by Chumphorn