31 August 2021 in Jurnal Elektronika dan Telekomunikasi
SSSulistyaningsih Sulistyaningsih
Sulistyaningsih Sulistyaningsih
This paper presents the development of vision-based robotic arm manipulator control by applying Proportional Derivative-Pseudoinverse Jacobian (PD-PIJ) kinematics and Denavit Hartenberg forward kinematics. The task of sorting objects based on color is carried out to observe error propagation in the implementation of manipulator on real system. The objects image captured by the digital camera were processed based on HSV-color model and the centroid coordinate of each object detected were calculated. These coordinates are end effector position target to pick each object and were placed to the right position based on its color. Based on the end effector position target, PD-PIJ inverse kinematics method was used to determine the right angle of each joint of manipulator links. The angles found by PD-PIJ is the input of DH forward kinematics. The process was repeated until the square end effector reached the target. The experiment of model and implementation to actual manipulator were analyzed using Probability Density Function (PDF) and Weibull Probability Distribution. The result shows that the manipulator navigation system had a good performance. The real implementation of color sorting task on manipulator shows the probability of success rate cm is 94.46% for euclidian distance error less than 1.2 cm.
1 May 2022 in Journal of Robotics and Control (JRC)
The Nutrient Film Technique (NFT) method is one of the most popular hydroponic cultivation methods. This method has advantages such as easier maintenance, faster and optimal plant growth, better use of fertilizers, and less deposition. The disadvantages of NFT include the consumption of electrical power and the faster spread of disease. Therefore, NFT requires a good nutrient control and monitoring system to save electricity and achieve optimal growth and resistance to pests and diseases. In this study, a nutrient control was designed with indicators of pH and TDS levels and equipped with an Internet of Things (IoT) based monitoring system. The control system used is the Mamdani Fuzzy Inference System. The output of the system is the active time of the pH Up, pH Down, and AB Mix nutrient pumps, which aim to normalize the pH and TDS of nutrient liquids. The experimental results show that one to three control steps are needed to normalize pH. One control step has a response time of 60 seconds, and it can prevent pH Up and pH Down oscillations. As for TDS control, the prediction of AB mix pump active time works accurately, and TDS levels can be normalized in one control step. Overall, based on surface control, simulations, and real experimental data, it is indicated that the control system operates very well and can normalize pH and TDS to the desired normal standard.
1 September 2023 in IAES International Journal of Artificial Intelligence (IJ-AI)
Artificial intelligence technology has dramatically improved the quality of services for human needs, one of which is technology to improve the quality of services for the blind and visually impaired, particularly technology that can help them understand visual sights to facilitate navigation in their daily lives. This study developed an image captioning model to aid the blind and visually impaired in outdoor navigation. The image captioning model employs the encoder-decoder method, with the convolutional neural network (CNN) feature extraction and attention layer as encoders and the long short-term memory (LSTM) as decoders. ResNet101 and ResNet152 are used in the encoder to extract image features. The results of the extraction and caption are forwarded to the attention layer and the LSTM network. The attention layer uses the Bahdanau attention mechanism. The accuracy of the model is calculated using the bilingual evaluation understudy score (BLEU), metric for evaluation of translation with explicit ordering (METEOR) and recall-oriented understudy for gisting evaluation-longest common subsequence (ROUGE-L). ResNet101 performed the best on BLEU-4, scoring 91.811% and 94.0337% in the METEOR evaluation. The captioning results show that the model is quite successful in displaying a simple caption that is suitable for each image.
15 October 2023 in International Journal of Advances in Intelligent Informatics
.This research developed a pest detection model for Indonesian red chili pepper based on fine-tuned YOLOv5. Indonesian red chili pepper is the third largest vegetable commodity produced in Indonesia. Pest attacks disrupt the quantity and quality of crop yields. To control pests effectively, it is necessary to detect the type of pest correctly. A viable solution is to leverage computer vision and deep learning technologies. However, no previous studies have developed a pest detection model for Indonesian red chili pepper based on this technology. YOLOv5 is a variant of the YOLO object detection algorithm, which has major advantages in terms of computation cost and execution speed. The dataset comprises 4,994 image files collected from a chili plantation in Bengkulu province, Indonesia, covering 4 different classes and a total of 10,683 pests. The image is 1216 x1216 px with the smallest, largest, and average object dimensions of 2%, 35%, and 4% of the image dimensions. The training model used is fine-tuning YOLOv5s with variations of patience as an early stop parameter of 100, 200, and 300. The evaluation of the trained model is based on train loss, validation loss, and mAP@0.5:0.95, the best-trained model is the 445th epoch on patience 100 with the best confidence value of 0.321 and the highest TF1 of 0.74. From the best-trained model testing on the test dataset, the mAP@0.5 performance for all classes is 81.3%. The model not only detected large pests but was also able to detect objects that were small in size compared to the image size. The best-trained model's best mAP@0.5 performance and speed are 82.6% and 20 ms/image, or 50 fps on NVIDIA P100 GPU.
1 March 2023 in Computer Science and Information Technologies
Sea turtles are reptiles listed on the international union for conservation of nature (IUCN) red list of threatened species and the convention on international trade in endangered species of wild fauna and flora (CITES) Appendix I as species threatened with extinction. Sea turtles are nearly extinct due to natural predators and people who are frequently incorrect or even ignorant in determining which turtles should not be caught. The aim of this study was to develop a classification system to help classify sea turtle species. Therefore, the ensemble deep learning of convolutional neural network (CNN) method based on transfer learning is proposed for the classification of turtle species found in coastal communities. In this case, there are five well-known CNN models (VGG-16, ResNet-50, ResNet-152, Inception-V3, and DenseNet201). Among the five different models, the three most successful were selected for the ensemble method. The final result is obtained by combining the predictions of the CNN model with the ensemble method during the test. The evaluation result shows that the VGG16 - DenseNet201 ensemble is the best ensemble model, with accuracy, precision, recall, and F1-Score values of 0.74, 0.75, 0.74, and 0.76, respectively. This result also shows that this ensemble model outperforms the original model.
30 May 2022 in Jurnal Amplifier : Jurnal Ilmiah Bidang Teknik Elektro dan Komputer
An egg incubator is a tool that helps the process of hatching eggs using an electric heater and is equipped with an egg rack that functions to evenly distribute the heat in the incubator. Good temperature control in the hatching process is something that greatly affects the hatching results. In this study, an egg incubator temperature control system was designed using the PID method with the Ziegler Nichols Open Loop Step Response tuning method. The control system is specifically for free-range chicken eggs which require a normal temperature of 37 °C-39 °C. The main control components are the microcontroller, the incandescent lamp heater, and the DHT22 temperature sensor. The open loop test shows a time delay of 20 seconds and a time constant of 385 seconds, so with the Ziegler Nichols open loop tuning method, the values of Kp = 23.1, Ki = 40, and Kd = 10. The test results show that the PID controller can control the temperature properly. In testing the hatching process within 21 days, the temperature control worked well, and the effect of changes in day and night temperature did not significantly affect the performance of the PID controller.