Pratama, Adi Wira / AWP (2025) Pemodelan Prediksi Kelelahan Kognitif Aktivitas Berkendara Menggunakan Driving Simulator dengan Explainable Machine Learning. Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5010211058-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (10MB) | Request a copy |
Abstract
Kecelakaan lalu lintas di Indonesia terus meningkat, dengan faktor manusia sebagai penyebab utama. Kelelahan kognitif menjadi salah satu faktor utama yang menurunkan pengambilan keputusan serta respon pengendara, sehingga meningkatkan urgensi pengelolaan kelelahan untuk pengendara. Berbagai studi telah menerapkan machine learning dan deep learning untuk mendeteksi kelelahan, namun keterbatasan interpretabilitas menjadi kendala implementasi. Penelitian ini bertujuan untuk mengembangkan model prediksi kelelahan kognitif berbasis explainable machine learning dengan menggunakan algoritma Kolmogorov–Arnold Network (KAN). Data diperoleh melalui eksperimen dengan desain cross-sectional menggunakan simulator berkendara. Model dibangun dengan mempertimbangkan variabel demografik serta beban kognitif yang dialami partisipan selama simulasi yang diukur menggunakan software OnDrive PVT yang mengukur waktu reaksi selama peserta berkendara. Kuesioner NASA-TLX serta kuesioner Swedish Occupational Fatigue Index (SOFI) digunakan sebagai instrumen pengukuran kelelahan utama. Penelitian ini berhasil mengembangkan model prediksi yang akurat dan interpretabel menggunakan KAN dalam dua bentuk yaitu Feed Forward Network (FFN) dan Recurrent Neural Network (RNN), dengan performa mendekati Random Forest, FFN umum, serta RNN umum. Penelitian ini menunjukkan bahwa KAN mampu memodelkan data sekuensial kompleks sekaligus memberikan interpretabilitas yang dibutuhkan dalam pengambilan keputusan di berbagai sektor.
=====================================================================================================================================
Traffic accidents in Indonesia continue to rise, with human factors being the primary cause. Cognitive fatigue is one of the key contributors to impaired decision-making and slower driver responses, thereby increasing the urgency for effective fatigue management among drivers. Various studies have applied machine learning and deep learning to detect fatigue, but limitations in interpretability have hindered practical implementation. This study aims to develop a cognitive fatigue prediction model based on explainable machine learning using the Kolmogorov–Arnold Network (KAN) algorithm. Data were collected through a cross-sectional experimental design using a driving simulator. The model was constructed by taking into account demographic variables and the cognitive load experienced by participants during the simulation, measured using the OnDrive PVT software that captures reaction time while driving. The NASA-TLX questionnaire and the Swedish Occupational Fatigue Index (SOFI) were used as the primary instruments for fatigue assessment. This study successfully developed an accurate and interpretable prediction model using KAN in two architectures: Feed Forward Network (FFN) and Recurrent Neural Network (RNN), with performance comparable to Random Forest, conventional FFN, and conventional RNN models. The results demonstrate that KAN is capable of modeling complex sequential data while also providing the interpretability needed for informed decision-making across various sectors.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Explainable Machine Learning, Kelelahan Kognitif, Simulator Berkendara |
Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. T Technology > T Technology (General) > T55 Industrial Safety |
Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26201-(S1) Undergraduate Thesis |
Depositing User: | Adi Wira Pratama |
Date Deposited: | 30 Jul 2025 08:37 |
Last Modified: | 30 Jul 2025 08:37 |
URI: | http://repository.its.ac.id/id/eprint/123751 |
Actions (login required)
![]() |
View Item |