Yazar "Oguz, Yuksel" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • [ X ]
    Öğe
    A Comparison of Multi-Layer Perceptron and Inverse Kinematic for RRR Robotic Arm
    (Gazi Univ, 2024) Aysal, Faruk Emre; Celik, Ibrahim; Cengiz, Enes; Oguz, Yuksel
    In this study, the position control simulation of a 3 Degree of Freedom (3DOF) robot arm was compared with machine learning and inverse kinematic analysis separately. The considered robot arm is designed in RRR pattern. In the inverse kinematic analysis of the robot arm, the geometric approach and the analytical approach are used together. Multi-Layer Perceptron (MLP) was used as a machine learning method. Some of the coordinate data that the robot arm can reach in the working space are selected and the MLP model is trained with these data. When training was done with MLP machine learning method, the correlation coefficient (R2) was obtained as 1. Coordinates of 3 different geometric models (helix, star and daisy) that can be included in the working space are used as test data of the MLP model. These tests are simulated in 3D in MATLAB environment. The simulation results were compared with the inverse kinematics analysis data. As a result, Mean Relative Error (MRE) values for helix, star and daisy shapes were calculated as 0.0007, 0.0033 and 0.0011, respectively, in the tests performed. Mean Squared Error (MSE) values were obtained as 0.0034, 0.0065 and 0.0040, respectively. This confirms that the proposed MLP model can operate this system at the desired stability.
  • [ X ]
    Öğe
    Cherry Tree Detection with Deep Learning
    (Institute of Electrical and Electronics Engineers Inc., 2022) Ozer, Tolga; Akdogan, Cemalettin; Cengiz, Enes; Kelek, Muhammed Mustafa; Yildirim, Kasim; Oguz, Yuksel; Akkoc, Hasan
    In recent years, many studies have been conducted on artificial intelligence. Artificial-intelligence-based applications appear in many fields, such as the defense industry, agriculture, transportation, and health. Food production and supply are critical with the increase in the world population and global warming. For this reason, it is seen that various artificial-intelligence-based applications in agriculture are increasing today. In this study, artificial-intelligence-based Cherry tree detection was carried out using the deep learning method. A DJI Mavic air drone collected images of cherry trees in the Afyonkarahisar. A cherry tree dataset was created using these images. The training was carried out with YOLOv5m, YOLOv5s, and YOLOv5x models. As a result of the training, F1 scores of 94.20%, 98.0%, and 95.9% were obtained. The experimental results obtained as a result of the training of the models were shared comparatively. © 2022 IEEE.