JITCE (Journal of Information Technology and Computer Engineering)

Control System Strategy for Ring Thrower Robot Based on PID-CSA for ABU Robocon 2023

2024-03-31T02:03:26-04:00

Thrower Robot need to maintain system stability in carrying out their tasks, which require accuracy and stability while throwing the ring in different zones and have different distances and height, so the thrower robot needs to maintain the stability of the system in order to reach target properly. Maintain the stability of the thrower is important because of the physical task like throwing the ring. However, disturbance from external systems can affect the accuracy which can reduce the performance of the robot while performing their task. Therefore, system needs stable accuracy in performing the tasks despite interference. The control system is used to maintain acceleration and elevation in the process of throwing the ring so that it can reach the specified target. The implemented system uses Proportional, Integral and Derivative (PID) control based on the Cuckoo Search Algorithm (CSA). Function of PID control is to maintain a constant position at a certain target and CSA is used to simplify PID control tunning when it has some parameter modifications. Therefore, combination of PID-CSA is applied for this system to produce a control system that aims to maintain stability and reduce disturbances contained in the ring throwing robot based on manipulation. From the result obtained, the PID-CSA method has a better level of stability because it can reduce the percentage value of the error which produced by PID-TE by showing the percentage value of distance error up to 0.68% and value of angle error up to 2.39%.

Nudibranch Suborders Classification based on Densely Connected Convolutional Networks

2024-03-31T02:03:26-04:00

Nudibranchs, often called sea slugs, are a group of soft-bodied marine gastropod mollusks that shed their shells after their larval stage. With their body structure that is very similar between one suborder and another, sometimes it is hard to tell apart the suborder of a nudibranch. In this work, we make an Image Classification model for determining the suborder of a nudibranch using deep learning algorithms DenseNet and EfficientNet. The experiment is conducted using Google Colaboratory environment. For DenseNet, we use 121, 169, and 201 layers; meanwhile, we only use the baseline algorithm for EfficientNet. The dataset for research is randomly taken from marine fauna forums on the internet. DenseNet with 201 layers shows a better generalization than other classifiers (accuracy of DenseNet 121, 169, 201, and baseline EfficientNet, respectively 53%, 41%, 73%, and 47%). The research produces a decent system for classifying the suborder of the Nudibranch. Usage of image recognition or background blurring systems in future research can improve the system's accuracy.

Deep Learning-Based Dzongkha Handwritten Digit Classification

2024-03-31T02:03:25-04:00

In computer vision applications, pattern recognition is one of the important fields in artificial intelligence. With the advancement in deep learning technology, many machine learning algorithms were developed to tackle the problem of pattern recognition. The purpose of conducting the research is to create the first-ever Dzongkha handwritten digit dataset and develop a model to classify the digit. In the study, the 3 layer set of CONV → ReLU → POOL, followed by a fully connected layer, dropout layer, and softmax function were used to train the digit. In the dataset, each class (0-9) contains 1500 images which are split into train, validation, and test sets: 70:20:10. The model was trained on three different image dimensions: 28 by 28, 32 by 32, and 64 by 64. Compared to image dimensions 28 by 28 and 32 by 32, 64 by 64 gave the highest train, validation, and test accuracy of 98.66%, 98.9%, and 99.13% respectively. In the future, the sample of digits needs to be increased and use the transfer learning concept to train the model.

The Evaluation of LSB Steganography on Image File Using 3DES and MD5 Key

2024-03-31T02:03:26-04:00

Information security is paramount for individuals, companies, and governments. Threats to data confidentiality are increasingly complex, demanding strong protection. Therefore, cryptography and steganography play pivotal roles. This study proposes the utilization of LSB (Least Significant Bit) steganography on image files employing the 3DES (Triple Data Encryption Standard) algorithm. The aim is to facilitate secure transmission and reception of data, including confidential messages, within digital information media. The research methodology involves implementing 3DES + LSB using Image Citra and innovating 3DES + MD5 Hash in .txt files. The results and discussions described include, among others, Pseudocode, Cryptographic Testing, and Steganography Testing. Based on the results of program analysis and testing, it can be concluded that the more messages that are inserted in the image, the more pixel differences there are in the stego image. The more colors in the image to which the message will be inserted, the more pixel differences in the stego image will be. The images that stego objects can present are only images with .png and .jpeg extensions. Testing from the fidelity aspect, the average PSNR obtained is 66,365, meaning that the stego image quality is very good. Testing from the recovery aspect, from 4 tested stego images, showed that messages can be extracted again. Testing of the robustness spec using two attack techniques, namely rotation, and robustness, shows that the message cannot be extracted from the image. Testers of the computation time, from testing 1-1000 characters, show the average time required for computation is about 0.798 seconds.

Design of a Drowsiness Prevention Helmet with Vibration and IoT-Based Theft Detection Alarms

2024-03-31T02:03:26-04:00

Ensuring safety while riding a motorbike is an imperative task. Currently, safety products such as helmets have the capability to provide protection to users without the additional feature of issuing warnings. Consequently, a preemptive alert system is developed to offer timely notifications to drivers. The experimental setup involves the utilization of a Max30100 sensor that is linked to a microcontroller and integrated into a helmet. The objective of this final project is to offer a timely alert to the rider and utilize the Max30100 sensor for pulse detection in order to ascertain the normalcy of the rider's pulse. In instances where the rider encounters tiredness and fatigue, it is common for the pulse intensity to exhibit a reduction. The Blynk application presents the detection pulse findings on the smartphone screen, while the buzzer on the helmet will activate in response to vibrations and sounds once the pulse has diminished. Based on testing, the average pulse rate on quiet road conditions is 78.58 BPM. On busy road conditions, the average pulse rate is 73.25 BPM. While in traffic conditions, the average pulse rate is 73.5 BPM. The helmet theft detector uses a Sharp GP2Y0A21 sensor that can only detect object distances up to 10 cm.