(#51) Review: Obstacle Detection and Localization in an Autonomous Vehicle

Abstract

This paper analyses the revolutionary impact of artificial intelligence (AI) technologies on the fast-changing landscape of vehicle automation within the automobile industry. AI software for automobiles is expected to generate USD 6.6 billion by 2025, fuelling a revolutionary transformation, particularly in the realm of autonomous driving systems. Object detection, a cornerstone of selfdriving vehicles, is examined through the lenses of two-stage and single-stage approaches, stressing its vital role in allowing vehicles to observe and classify their surroundings. The analysis goes deeply into the complexities of obstacle avoidance, emphasizing its critical role in real-time path planning and assuring human and vehicle safety. Object localization, a critical component of this process, requires precise distance determination and justified coordinates to overcome obstacles given by changing environmental conditions. Current research highlights the dominance of deep learning algorithms, with the YOLO approach and several R-CNN variations taking the lead in object identification and avoidance systems. The problems of sensor integration are examined, with a focus on cameras and LIDARs, which are judged appropriate for 3D modeling. The review also addresses sensor position issues and provides an ideal solution that uses cameras for object recognition and LIDARs for depth perception. The literature study continues by identifying relevant studies with approaches that are closely matched with the suggested methodology, underlining the importance of continuous research to accelerate developments in autonomous driving capabilities.

Description

Citation

Weerasinghe, K.M.R.; Lakmal, H.K.I.S. (2024) (#51) Review: Obstacle Detection and Localization in an Autonomous Vehicle, Technology Innovations & Digitalization Accompanying Creativity (TIDAC), NSBM Green University, Sri Lanka. P.105-109

Endorsement

Review

Supplemented By

Referenced By