Sürdürülebilir Kalkınma Amaçları
Yayınlar
- MakaleAçık Erişim2026Scopus SJRQ4Enhanced control of a two-wheeled gyroscopic combat robot with variable CMG disk speeds using LQR-Based neural network
Oğuz, Yasin; Ünker, Faruk; Cetin, Gokhan
This study proposes a hybrid control approach to improve the stability and adaptability of a two-wheeled gyroscopic combat robot by integrating deep learning with Linear Quadratic Regulator (LQR) control. Due to their inherently unstable nature, two-wheeled robots face challenges maintaining balance, especially when Control Moment Gyroscope (CMG) disk speeds vary, which significantly affect system dynamics. While LQR provides effective stabilization by minimizing a cost function, its performance degrades when disk speeds change. To overcome this limitation, a Neural Network (NN) was trained on 42001 precomputed LQR gain sets optimized for CMG disk speeds ranging from 800 to 5000 rad/s. The trained NN predicts appropriate coefficients in real time, enabling adaptive control under dynamic operating conditions. Simulation results demonstrate that the hybrid NN-LQR controller maintains stability under recoil forces up to 3000 N, where the conventional LQR loses balance about 4–6 seconds. Furthermore, the proposed controller reduces energy consumption by 21.6% during steady-state operation within the first 5 seconds and 10.6% under recoil conditions within the first 3 seconds compared to classical LQR. Although developed for a combat robot, the proposed framework offers a scalable solution for broader robotic applications such as search and rescue, surveillance, and autonomous navigation in complex environments, contributing to the advancement of intelligent control in robotics.
- MakaleAçık Erişim2025Advanced adaptive control strategies application for speed regulation of a 12V small dc geared motor
Cetin, Gokhan
This paper investigates the implementation and performance of adaptive control techniques for a 12V small geared DC motor characterized by modeling errors and input disturbances. This paper discusses the following two primary approaches: Adaptive Radial Basis Function Neural Network (ARBFNN) Controllers and Model Reference Adaptive Control (MRAC). In model uncertainty, MRAC and ARBFNN outperformed the simple Proportional-Integral (PI) controller. The study is further expanded to involve Robust MRAC and Adaptive Sliding Mode Radial Basis Function Neural Network (ASRBFNN) Controllers to counter the compounded effects of model uncertainty and input disturbances. The versions of the robust controllers performed better than the conventional PI controller in cases involving both uncertainties and disturbances. Implementations were done on a 12V geared DC motor testbed with an Arduino microcontroller and MATLAB's System Identification Toolbox. The results from simulations and experimental applications highlight the greater flexibility and disturbance rejection capability of the developed advanced adaptive control schemes, making them perform better than standard PI controllers under challenging conditions.
- BildiriYalnız künye2023Pedestrian Inertial Navigation with Multi-Head CNN
Cetin, Gokhan; Kucuk, Mehmet Ali; Koroglu, Muhammed Taha
Traditional pedestrian dead-reckoning (DR) methods utilizing inertial measurement units (IMU) have been successfully employed in GPS-denied environments over the past two decades. However, despite continuous advancements in wearable IMU technology, the long-term solution still suffers from the issue of drift. In light of this, Deep Learning (DL) has emerged as a potential candidate for inertial navigation solutions, leveraging its powerful non-linear regression capabilities. Recent research has demonstrated that DL models trained on public datasets can effectively serve as end-to-end solutions. Most studies focus on smartphones, where the IMU is typically situated at the torso, and generally work with transformed data represented in the navigation coordinate frame to estimate pedestrian velocity and heading change at rates lower than the IMU sampling rate. In contrast, this study uses raw foot-mounted IMU data as input to a multi-headed structure consisting of independent convolutional neural networks (CNNs) and targets displacement directly and heading change. The proposed model is trained on a publicly available foot-mounted IMU dataset where pseudo ground-truth (GT) data is generated heuristically to provide supervision for the learning process. The results indicate that the Multi-Head CNN (MHCNN) model is capable of producing pedestrian trajectories at the IMU sampling rate, with positioning errors of less than 2m throughout the duration of the walks. The code for the trained network can be accessed in the associated repository, i.e., github.com/mtkoroglu/PIN-MHCNN.
AtıfScopus1
İncele
