Article Per Year (5 Year)
p-Index From 2021 - 2026
0.702
P-Index
This Author published in this journals
All Journal IAES International Journal of Artificial Intelligence (IJ-AI) Journal of Robotics and Control (JRC) GEOGRAPHIA: Jurnal Pendidikan dan Penelitian Geografi
Aulia Siti Aisjah
Institut Teknologi Sepuluh Nopember Surabaya
Aerospace Engineering Computer Science & IT Control & Systems Engineering Earth & Planetary Sciences Education Electrical & Electronics Engineering Engineering Environmental Science Mechanical Engineering Social Sciences Other

Published : 4 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 4 Documents
Search

 1 
Detecting road damage utilizing retinanet and mobilenet models on edge devices Mahmudah, Haniah; Aisjah, Aulia Siti; Arifin, Syamsul; Prastyanto, Catur Arif
IAES International Journal of Artificial Intelligence (IJ-AI) Vol 14, No 2: April 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijai.v14.i2.pp1430-1440

Abstract

A particular form of road digitalization produces a system that detects road damage automatically and in real time, employing the device to detect road damage as an edge device. The application of RetinaNet152 and MobileNetV2 models for road damage detection on edge devices necessitates a trade-off between high system performance and efficiency. Currently, edge devices have limited storage. In this paper, we explore how tuning hyperparameters with batch size and several optimizers improves system performance on RetinaNet152 and MobileNet models, as well as how they are implemented on edge devices. After tuning hyperparameters in the batch size of the optimizer, the Adam optimizer displayed enhanced performance with mean average precision (mAP), average recall (AR), and F1-score. This implies a positive impact on overall model performance. The MobileNetV2 model's hyperparameter tuning technique significantly improves performance, resulting in faster inference times and overall system performance. This demonstrates that the MobileNetV2 model could be used directly on edge devices to identify road damage. However, the RetinaNet152 model has a lower inference time, which cannot be deployed directly to edge devices. The RetinaNet152 model can be deployed on edge devices; however, a technique for speeding up inference time is essential.
Q-RCR: A Modular Framework for Collision-Free Multi-Package Transfer on Four-Wheeled Omnidirectional Conveyor Systems Kautsar, Syamsiar; Aisjah, Aulia Siti; Arifin, Syamsul; Syai'in, Mat
Journal of Robotics and Control (JRC) Vol. 6 No. 4 (2025)
Publisher : Universitas Muhammadiyah Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.18196/jrc.v6i4.26050

Abstract

Modern logistics systems increasingly require high flexibility in handling simultaneous package transfers in compact, dynamic environments without collisions. Improper handling of multi-package transfers in omnidirectional conveyor systems can lead to deadlocks, congestion, or delivery delays, particularly in grid-based environments where routing complexity increases with package variability and layout density. This research addresses these challenges by introducing Q-RCR, a modular Q-Learning-based framework with Rule-Based Conflict Resolution (RCR) for intelligent path planning and collision handling in Four-Wheeled Omnidirectional Cellular Conveyor (FOCC) systems. The research contribution is decoupling path learning and collision handling, enabling independent agent training while minimizing computational burden and improving convergence in multi-agent scenarios. The proposed Q-RCR framework integrates Q-Learning for route optimization with a rule-based conflict resolution module, applying four adaptive strategies: Sequential Transfer, Insert Path, Reroute, and Hybrid. The method is implemented in a grid-based FOCC environment, supporting eight-directional movement and handling various package sizes. Experiments were conducted in four scenarios with grid dimensions ranging from 8×11 to 12×12 and involving up to four simultaneous packages. Results show that Q-RCR consistently outperforms Double Q-Learning, RRT, and A* regarding delivery time, path smoothness, and the number of activated cells. The hybrid mode demonstrated the most effectiveness in handling frequent collisions and maintaining operational flow continuity. The proposed framework demonstrates strong adaptability, scalability, and responsiveness, offering a practical and intelligent solution for real-time multi-package coordination in flexible manufacturing and warehouse automation environments.
Earthquake clustering analysis in North Sumatra region based on double-difference relocation Purba, Rini Kumala Sari; Aisjah, Aulia Siti; Arifin, Syamsul; Nugroho, Hendro
GEOGRAPHIA : Jurnal Pendidikan dan Penelitian Geografi Vol. 6 No. 2 (2025): Desember
Publisher : Jurusan Pendidikan Geografi Universitas Negeri Manado

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.53682/e7n0yj02

Abstract

North Sumatra is an active tectonic zone influenced by the complex interaction between the subduction of the Indo-Australian plate and the movement of the Sumatra Fault. Accurate determination of earthquake hypocenter locations is crucial for seismotectonic modeling and disaster mitigation. This study aims to improve the precision of earthquake hypocenter location in this region through the implementation of the Double-Difference (HypoDD) relocation method and to analyze clustering based on the spatial distribution of earthquakes to identify active fault segments. The earthquake data used came from local seismic station catalogs during the period 2008–2024. The application of the HypoDD method significantly reduced the Root Mean Square (RMS) value of the average seismic phase arrival residual from 0.79373 to 0.33888, indicating an increase in the accuracy of the hypocenter location. Earthquake clustering analysis identified a total of 10 clusters that showed a strong correlation with the surrounding geological structure. The shallow earthquake cluster in the Tarutung region was dominated by a strike-slip fault mechanism, which definitively confirmed that the Renun Segment is an active segment of the Sumatra Fault with intense activity. In addition, a swarm pattern with low magnitude and very shallow depth (<15 km) was identified in the Toba Basin (Cluster 2), indicating the contribution of tectonic-magmatic processes. The cluster of medium- to deep-depth earthquakes (70–150 km) is strongly associated with subduction activity in the Sumatra intraslab megathrust zone. Overall, this study successfully mapped the spatial distribution pattern of earthquake sources in greater detail, contributing significantly to the updating of earthquake hazard maps and the determination of active fault zones in North Sumatra.
Daily rainfall forecast based on multi-station observation data in Medan City Sinaga, Nora Valencia; Aisjah, Aulia Siti; Arifin, Syamsul; Saragih, Immanuel Jhonson Arizona
GEOGRAPHIA : Jurnal Pendidikan dan Penelitian Geografi Vol. 6 No. 2 (2025): Desember
Publisher : Jurusan Pendidikan Geografi Universitas Negeri Manado

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.53682/82y51t67

Abstract

This study develops a multi-horizon daily rainfall forecasting model using the Long Short-Term Memory (LSTM) deep learning method, based on multi-station Automatic Weather Station (AWS) data in Medan City. Ten-minute AWS data from multiple stations (2021–2024) were merged and time-synchronized (UTC), followed by a quality control process including physical range checks, rate-of-change filtering, inter-variable consistency checks, and spike detection. Missing values were addressed using linear interpolation for short gaps and Multiple Imputation by Chained Equations (MICE) for longer gaps. Predictor features were constructed from weather parameters (temperature, humidity, pressure, wind, radiation), aggregated to an hourly scale, and reshaped into input time windows for LSTM. A two-layer LSTM model (128–64 units, 0.3 dropout, Adam optimizer) was trained to predict daily rainfall up to five days ahead. Evaluation metrics, including RMSE, MAE, POD, FAR, and CSI (with rainfall threshold ≥1 mm/day), indicated strong model performance: for instance, RMSE was below 10 mm/day for 1–3 day horizons, with POD above 0.80 and FAR below 0.20. The LSTM model outperformed conventional statistical models, yielding an accuracy improvement of approximately 30–40%. These findings highlight the potential of high-resolution AWS-based automatic forecasting systems to support hydrometeorological disaster mitigation in tropical urban areas.
Co-Authors Catur Arif Prastyanto, Catur Arif Hendro Nugroho, Hendro Kautsar, Syamsiar Mahmudah, Haniah Purba, Rini Kumala Sari Saragih, Immanuel Jhonson Arizona Sinaga, Nora Valencia Syai'in, Mat Syamsul Arifin