ECTI Transactions on Computer and Information Technology (ECTI-CIT)

Ladybug: An Automated Cultivation Robot for Addressing the Manpower Shortage in the Agricultural Industry

Apirak Tooltham — 2024-03-30

The agricultural sector is projected to need more labor as a result of declining interest in careers within this domain. Despite the escalating demand for agricultural goods, previous endeavors to mitigate this challenge through the deployment of robotic prototypes have encountered hindrances such as issues pertaining to automation, adaptability to varying tasks, and the financial burdens associated with development. To address this exigency, we have developed an automated cultivation robot utilizing advancements in the Internet of Things (IoT), Image Processing, and artificial Intelligence (AI) for seeding in pots. The robot demonstrates the capacity to sow seeds in 257 pots per hour, accomplish a mission within 12.53 minutes, traverse at a velocity of 360 meters per hour, and seed pots at a rate of 13.37 seconds per pot. It possesses an operational duration of approximately two hours, completing nine cycles and seeding 486 pots on a single charge. Notably, the robot exhibits a mission success rate of 1.00 and a seeding accuracy 0.78. Moreover, it features an adaptable workspace and a lightweight frame weighing 20 kg, rendering it a cost-effective solution for mass production.

An IoT-based Multi-sensory Intelligent Device for Bedridden Elderly Monitoring

Worrakrit Wudhiphan — 2024-03-30

A significant responsibility of elderly caregivers is monitoring the health condition of the elderly. Health monitoring can become increasingly difficult for caregivers of bedridden older adults since they cannot lie in the same position for periods longer than two hours. Therefore, we used an artificial intelligence to alert caregivers and alleviating their workload. This work aims to develop a system to support the care of bedridden older adults using the SensorTag CC2650STK as a motion sensor. We used accelerometers and gyroscopes to generate the model for analyzing the lying position of older adults. The system can help caregivers by sending notifications when older adults have been lying in the same position for too long. We dened the lying position into four classes: sit, left, right, and back. Three machine learning models (K-NN, Decision Tree, and Naïve Bayes) were generated and evaluated in our work. We found that the decision tree could achieve the best classification results among these ML models, obtaining scores of 0.98, 0.97, and 0.97 for precision, recall, and F1 scores, respectively.

Enhancing Password Storage Through the Integration of Cryptarithmetic Techniques and Hash Functions

Jakkapong Polpong — 2024-04-06

The utilization of internet-based applications oers numerous benefits and significantly enhances the daily lives of individuals. To access these systems, users must provide unique username and password credentials, verifying their identity as the legitimate owners of the data. However, there are instances where unauthorized individuals, including hackers, gain access to these systems by illegitimately exploiting these credentials. This study aims to enhance the system's efficiency by introducing a novel algorithm named Cryptarithmetic_shield. This algorithm combines the Cryptarithmetic technique with a hash function to create a secure encryption system. The research results indicate that password data encrypted using the Cryptarithmetic_shield algorithm, in conjunction with the hash functions MD5, SHA1, SHA256, SHA512, CRC32, and RIPEMD160, effectively prevents decoding from Dictionary attack and Rainbow Table Attack methods, achieving up to 100% effectiveness.

Multi-Task in Autonomous Driving through RDNet18-CA with LiSHTL-S Loss Function

shang shi — 2024-04-06

Most current autonomous driving research focuses on single-task or dual-task methods. We propose to combine road tracking, obstacle avoidance, traffic sign recognition, and traffic light recognition in a single multi-task framework. Additionally, we validate it using a scale model car to confirm its viability in a semi-physical environment. We propose a novel framework, RDNet18-CA, designed to reduce the training requirements associated with enormous full-scene datasets. These massive full-scene datasets are utilized in the autonomous driving systems of companies such as Google and Tesla. Thus, our framework performs well with small training datasets and can function in unseen scenarios to a certain degree. Additionally, we present an innovative loss function, LiSHTL-S, that exhibits adaptivity. This allows the LiSHTL-S loss function to be dynamically modified based on the properties of the train data and the state of the model throughout the training phase, eliminating the requirement for intense manual parameter tuning. Lastly, we present a new traffic light design concept called the traffic board to enhance its resistance to lighting noise, making it more adaptable for autonomous driving. With these innovations in mind, our method outperforms existing methods in multiple areas.

Enhancing Text Summarization using Hybrid LSTM-GRU with Lingual Significance Relation-based Attention Mechanism Model

PL. Prabha — 2024-04-13

Text summarization is the process of summarizing the information of a large text into short, crisp, and concise text to analyze and extract the most imperative information from the given text. Therefore, different AI based techniques are used for summarizing the text. In order to achieve this, various AI techniques have been incorporated into the existing works. However, the prevailing methods lagged in delivering accurate text. Therefore, the proposed work employs a Hybrid LSTM-GRU (Long Short Term Model Gated Recurrent Unit) model with M-AM (Modified - Attention Mechanism). The dataset incorporated in the proposed model is amazon fine food review. Different pre-processing techniques remove unwanted and irrelevant text, such as tokenization, text cleaning, stop word removal, and stemming and lemmatization. The Proposed model employs hybrid LSTM-GRU with M-AM as it delivers faster and employs less memory consumption. Along with it, it has the potential to capture long-term dependencies as well. Further, M-AM incorporates lexical sequence measure and sentence context weight for delivering an effective model for text summarization. Therefore, the major contribution of the proposed work involves summarizing the text into a crisp and brief format for easy understanding. Finally, the performance of the proposed model is evaluated using different ROUGE, accuracy, and loss, in which ROUGE metrics obtained by the proposed model is 55.5.

Hybrid Approaches for Efficient Simulations of 3-Qubit Quantum Fourier Transform (QFT) Circuit Using Quick Quantum Circuit Simulation (QQCS)

Thea Mayen Malimban — 2024-04-13

The research devised efficient methods for simulating 3-qubit Quantum Fourier Transform (QFT) circuits using Quick Quantum Circuit Simulation (QQCS). The hybrid methodologies suggested as a solution for efficiently simulating the circuit involve the combination of decision diagrams and property exploitation techniques. This paper incorporated two methods based on decision diagrams: the reordering trick and decision diagram approximations, template-based optimization, and linear reversible circuit synthesis for property exploitation. The proposed approaches significantly improved and optimized quantum algorithms and hardware by aiming to simulate quantum circuits accurately and quickly. Simulations using QQCS proved the effectiveness of these strategies, which were then compared to the original circuit. The results yielded valuable insights into enhancing simulation efficiency while upholding circuit accuracy.

Dementia U-Care: Comprehensive Cognitive Screening Application for Seniors

Egkarin Watanyulertsakul — 2024-04-27

The prevalence of cognitive impairment increases with age, particularly impacting seniors as it advances to severe dementia. These conditions pose significant challenges for afflicted individuals and their caregivers, manifesting as profound impacts on daily life and imposing considerable emotional and financial burdens on families. Mild cognitive impairment (MCI) denotes an intermediate stage between normal cognitive function and dementia, signifying a decline in cognitive abilities while maintaining normal daily life activities. Identifying MCI early in seniors within the community is pivotal to preventing further cognitive decline.
In response to the challenge of traditional cognitive function assessments, which require trained healthcare professionals and take 20-30 minutes per case, we introduce "Dementia U-Care," an innovative app designed to assist community health workers in screening, monitoring, and collecting cognitive data. Accessible on mobile devices, it allows seniors to respond through drawing and writing, simplifying data collection compared to paper forms. Dementia U-Care streamlines preliminary assessments, empowering professionals and reducing fatigue and errors. This tool enables prompt screening, minimizing test-related stress, with an average testing time of 13.06 minutes, ranging from 9 to 20 minutes. The evaluation indicates high satisfaction with Dementia U-Care, with a mean score of 9.37±1.12. Users are generally pleased with its quality and user experience, demonstrating its effectiveness in meeting their needs and providing a positive experience.