Citation

Perumal PS, Sujasree M, Chavhan S, Gupta D, Mukthineni V, Shimgekar SR, Khanna A, Fortino G. Eng. Appl. Artif. Intell. 2021; 104: e104406.

Copyright

(Copyright © 2021, Elsevier Publishing)

DOI

10.1016/j.engappai.2021.104406

PMID

unavailable

Abstract

Emergence of communication technologies made the automotive industries across the globe to embrace Advanced Driver Assistance Systems (ADAS) by considerable investments to ensure accident-free travel, reduction of pollution, fuel conservation. ADAS achieves its goals by integrating complex subsystems such as obstacle avoidance, overtaking advice, lane changing assistance, planning shortest routes, parking assistance, automatic gear shifting, etc., using the emerging technologies. This article emphasizes the road safety aspect of the ADAS by exploring Crash Avoidance and Overtaking Advice (CAOA) subsystems. Existing studies have a noticeable lack of connectivity between various aspects of CAOA subsystems. This review deeply explores and connects CAOA subsystems like road geometries, road debris, obstacle avoidance algorithms powered by Artificial Intelligence (AI), overtaking advice systems, perception challenges of human drivers in various light and weather conditions, driver inattention and misjudgments, vehicle blind-spots, vehicle parameter analysis, performance of vision sensors, in-vehicle computers, driver-vehicle interactions, Vehicle to Infrastructure (V2I) technologies. This article emphasizes the three primary performance metrics of the ADAS, namely accuracy, response time and robustness. Finally, this article discusses a typical functional architecture and gaps identified in existing studies. This article is structured to assist like-minded researchers, who work on CAOA systems for road safety.

Language: en

Keywords

ADAS; Communication network; Connected autonomous vehicles; Crash avoidance; Deep learning models; Driver assistance; In-vehicle computers; Primary vision sensors; V2X