INTERNATIONAL SCIENTIFIC JOURNAL OF ENGINEERING AND TECHNOLOGY (ISJET)

Design and Development of a Wireless Functional Electrical Stimulation and m-Health Application for Foot Drop Using an IoT-Based Architecture

Sat, 15 Nov 2025 00:00:00 +0700

To enhance accessibility in foot drop rehabilitation, this study presents the design and development of a wireless functional electrical stimulation system integrated with a mobile health (m-Health) application and an ESP32-based IoT (Internet of Things) platform. The system comprises a stimulation node and a sensor node that communicate via Bluetooth Low Energy (BLE) for heel-strike-triggered stimulation. The stimulation node delivers symmetrical biphasic pulses to the peroneal nerve with adjustable parameters. A cloud-based backend using MQTT supports real-time logging and device management, while the m-Health application enables mode selection, parameter tuning, and usage tracking. Key hardware includes a 70V boost converter, programmable current limiter, and H-bridge pulse generator. Evaluations show reliable pulse output (400.3 us width, 2.1 us rise/fall), low BLE latency (6.16 ms), and accurate analog-to-digital converter readings. Results confirm the system's feasibility as a compact, portable solution for home-based rehabilitation, addressing limitations of traditional wired systems.

Enhancing Vehicle Routing with Time Windows Solutions via K-means Clustering: A Comparative Study of Elbow and Truck Utilization Methods

Kanokporn Boonjubut, Prat Boonsam, Sirichai Yodwangjai — Sat, 15 Nov 2025 00:00:00 +0700

The vehicle routing problem with time windows is important in optimizing logistics distribution. For VRPTW optimization, a strategy is used to classify and optimize routes using artificial intelligence methods. Therefore, an improved two-phase algorithm is required to find a solution. Namely, a customer group can be divided into several regions using the K-means algorithm in the first phase, and each region can be decomposed into smaller subgroups according to certain constraints. In the second phase, local search from OR-tools solves the routing problem. In this experiment, two different methods of determining the number of clusters, namely, the elbow method and the truck utilization method, are compared by experimenting with a total of 26 standard instances. The results show that the truck utilization ratio outperforms the elbow method for the K-means algorithm in terms of overall results. The results from this experiment can be highly beneficial for routing, particularly when handling huge amounts of data that need to be subdivided ahead.

Intelligent Mobile-Based Detection of Shrimp Weight Anomalies Using Random Forest Regression

Sat, 15 Nov 2025 00:00:00 +0700

Shrimp is one of the most widely consumed seafood items globally, yet consumers frequently encounter fraud, such as weight manipulation through adulteration injections, which poses significant health and economic risks. This research presents a practical system for detecting anomalies in shrimp weight. A cross-platform mobile application has been developed to classify shrimp as either normal or abnormal in weight. The application integrates a shrimp segmentation model, developed using Mask R-CNN, and a weight prediction model based on the random forest algorithm, utilizing features such as area, perimeter, length, and width of the shrimp image. The weight prediction model achieves a value of 0.821 and a Mean Absolute Error (MAE) of 1.786 grams, which is less than 10% of the average shrimp weight in the dataset. Final classification is performed by comparing the predicted weight with the actual weight, measured using a 7-segment digit recognition module. The developed mobile application represents a novel integration of machine learning with mobile technology to address both non-adulterated and adulterated shrimp scenarios. It offers a reliable, accessible tool for consumers to detect weight-based adulteration, thereby helping to mitigate health risks and economic losses in the seafood supply chain.

Low-Cost Digitization for Monitoring Manual Processes: A Case Study of Tray Cleaning in Semiconductor Manufacturing

Pathitta Timtong, Pornnapat Kwanboonya, Warut Pannakkong — Sat, 15 Nov 2025 00:00:00 +0700

The increasing global adoption of Industry 4.0 technologies has transformed many aspects of manufacturing through automation, data analytics, and smart systems. However, high implementation costs often prevent labor-intensive processes, particularly in developing regions and smaller enterprises, from realizing these benefits. This study addresses that gap by proposing a low-cost digitization solution focused on tray cleaning in semiconductor manufacturing, a critical yet manually executed process. In collaboration with Sony Device Technology Thailand, an Excel VBA-based application was developed to automate real-time recording of tray movements and manpower data. The system incorporates inventory tracking, productivity monitoring, and an interactive dashboard that enhances operational visibility and eliminates the need for manual checks. Quantitative evaluation revealed a 50% reduction in input processing time and a 74.07% decrease in output processing time per transaction. Qualitative improvements include increased cross-functional collaboration, more transparent workforce evaluation, and data-driven decision-making. This study contributes a scalable, practical model for affordable digitization that integrates seamlessly with existing workflows, offering an effective path toward digital transformation in labor-intensive manufacturing environments.

Optimizing Tricycle (Tuk-Tuk) Suspension Systems Using Mathematical Modeling

Chaiyawoot Narintharangkul, Poom Jatunitanont — Sat, 15 Nov 2025 00:00:00 +0700

The use of tricycles in Thailand has declined due to mechanical issues, particularly with the engine and suspension systems. This research focuses on optimizing the suspension system to improve ride comfort. The study has three main objectives: 1) to analyze the vibration of the leaf spring suspension in two-section convertible tricycles, 2) to develop a suspension model that complies with ISO 2631-1 standards for vibration comfort, and 3) to compare vibration effects between the original and new suspension models using simulation software. The goal is to reduce vibrations and enhance comfort through mathematical modeling and parameter optimization. FFT and PSD analyses identified dominant vibration frequencies in the 4 to 8 Hz range, which correspond to human body resonance. By applying a band-stop filter and optimizing spring stiffness and damping coefficient, the damping ratio was adjusted to 0.3. This led to a significant reduction in RMS acceleration from 0.985 m/s² to 0.537 m/s², and peak acceleration dropped from 1.12 m/s² to 0.582 m/s², improving comfort from fairly uncomfortable to slightly uncomfortable.

The optimized suspension design significantly reduced vibrations in critical frequency ranges. The results from this approach can be applied to various vehicle types, offering potential for further development in the automotive industry, especially for vehicles sensitive to vibration.

Physical Interference Attacks on Autonomous Driving

Chuanxiang Bi, Jian Qu — Sat, 15 Nov 2025 00:00:00 +0700

Recent studies have revealed that there are serious security risks to autonomous driving, despite the notable advancements made by deep neural networks in this field. Simple sticker jamming has little experimental validation, despite recent proposals for physical attacks successfully implementing jamming in the real world and misleading autonomous driving recognition. This study focuses on the practicality of various sticker-based physical jammers, such as background noise, colorful stickers, smiley face stickers, and QR code stickers. To boost the study’s actual impartiality, we replace the genuine self-driving car in this work with a smart car that performs similar activities. We then utilize three models to train our dataset and carry out five sets of tests. Based on the results, it can be concluded that the QR code sticker has the most potential to interfere with the smart car. This interference causes the smart car’s accuracy in recognizing road signs to be between 30% and 40%, whereas the accuracy of the other interferences is over 50%. Furthermore, it demonstrated that, out of the three models, Resnet18 had the best anti-interference capability.