Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2408.04476

Impact Analysis of Data Drift Towards The Development of Safety-Critical Automotive System

Md Shahi Amran Hossain·Abu Shad Ahammed·Divya Prakash Biswas·Roman Obermaisser
View original source ↗Read source document ↗Export metadata ↓

Abstract

A significant part of contemporary research in autonomous vehicles is dedicated to the development of safety critical systems where state-of-the-art artificial intelligence (AI) algorithms, like computer vision (CV), can play a major role. Vision models have great potential for the real-time detection of numerous traffic signs and obstacles, which is essential to avoid accidents and protect human lives. Despite vast potential, computer vision-based systems have critical safety concerns too if the traffic condition drifts over time. This paper represents an analysis of how data drift can affect the performance of vision models in terms of traffic sign detection. The novelty in this research is provided through a YOLO-based fusion model that is trained with drifted data from the CARLA simulator and delivers a robust and enhanced performance in object detection. The enhanced model showed an average precision of 97.5\% compared to the 58.27\% precision of the original model. A detailed performance review of the original and fusion models is depicted in the paper, which promises to have a significant impact on safety-critical automotive systems.

Explore related subjects

math.LO

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Md Shahi Amran Hossain, Abu Shad Ahammed, Divya Prakash Biswas, Roman Obermaisser. 2024-08-07. Impact Analysis of Data Drift Towards The Development of Safety-Critical Automotive System. https://arxiv.org/abs/2408.04476

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Natural Term Logic

In this paper we develop a formal system called Natural Term Logic (NTL). NTL aims to represent key aspects of the logical and grammatical mechanisms of natural language as well as grammatical transformations which preserve core logical meaning. NTL can be seen as a refinement of the ideas of Quine's paper `Variables Explained Away' and the technical concepts introduced by Bealer and Zalta. NTL is more fine-grained than Bealer's first-order intensional logic (BL): there is a many-to-one correspondence $ν$ between NTL terms and closed BL terms as well as a canonical map $β$ which assigns to each closed BL term a corresponding NTL term. The map $ν$ can be seen as assigning a core logical content of the NTL term. We define a series of reductions on NTL terms which intuitivelyy speaking capture meaning-preserving syntactic transformations ( transformations which preserved the basic logical meaning of a term) and our main result is that each NTL term $T$ reduces to a unique normal term $N$. The reductions fall into the structural, predicative and pushing-in categories. Predicative reductions decompose NTL terms so that predication is only applied to a primitive term (such terms are called prenormal). A key ingredient in the proof is the fact that $βνN = N$ when $N$ is normal. This suggests that within NTL the normal form of a term expresses the core logical content of the term.

math.LO↗
arXiv2026

Hyper-hyperfiniteness and complexity

We show that if there exists a countable Borel equivalence relation which is hyper-hyperfinite but not hyperfinite, then the complexity of hyperfinite countable Borel equivalence relationsis as high as possible, namely, $Σ^1_2$-complete. We also establish an implication between the question of the effectivity of hyperfiniteness and its complexity.

math.LO↗
arXiv2026

Coordinate recognition: General theory, Groups, and other surprises

A class of structures \emph{recognizes coordinates} if any reduced product of structures from said class witnesses a certain kind of rigidity phenomenon. We provide several equivalent characterizations of this property. This property has (at least) two remarkable consequences, one set-theoretic and one model-theoretic, for reduced products of structures of the said class. First, under appropriate set-theoretic assumptions every isomorphism between such reduced products associated with the Fréchet ideal lifts (modulo a finite change) to an isomorphism between products of the original structures. Second, with an additional mild assumption, it implies a strong quantifier elimination result. Of note, we show that a class recognizes coordinates if and only if an individual formula witnesses a certain syntactic property. We also consider many concrete classes of structures and determine whether or not they recognize coordinates. We place heavy emphasis on well-known classes of groups, such as permutation groups, acylindircally hyperbolic groups, quasisimple groups, free products, and graph products, but we also discuss other classes of structures.

math.LO↗