SMART GREENHOUSE SYSTEM
Advisor: Asst. Prof. Tuba Ayhan
Team: Naci Gökberk TANDOĞAN, Samet Refik DERBENTLİ
Keywords: Greenhouse, Gas Sensor, Microcontroller, Arduino IoT Cloud, Lo-Ra
Abstract: The main idea of the project is to transmit carbon dioxide, humidity and temperature data in greenhouses to the user in a cheap, easy and fast way. In addition, it will be prevented from exceeding the upper limit of the optimal carbon dioxide level determined for plant growth. Blocking will be done with the help of a fan. The fan will activate when the maximum amount of carbon dioxide required for plant growth in the greenhouse is exceeded. Two separate systems will work on the project. The first system, which is the agent, will collect the data received from the greenhouse and send it to the host system. At the same time, it will decide when the fan will be activated and transmit the status of the fan to the host. The host part of the project will upload the data to a server. And finally, there will be an app and app that will display the data to the user. The app will be easy to understand and use. Our aim to produce cheaply will be achieved by imitating the data we receive from the high-precision infrared sensor through electro-chemical sensors and by choosing the cheapest of the other components used, which will provide the minimum expectation. Electro-chemical sensors to be used are cheaper than infrared sensors. The project will enable the greenhouse producer to monitor the carbon dioxide level regularly and uninterruptedly through the application and the website. Due to its cheapness, it is aimed to be easily accessible by every greenhouse producer and it is aimed to increase greenhouse efficiency throughout the country. The host part of the project will send sensor data to the agent part over Lo-Ra every 2-3 minutes. During this communication, data will be encrypted to prevent parasitic data
3D MAPPING SYSTEM FOR USE IN SELF-DRIVING VEHICLES
Advisor: Asst. Prof. Tuba Ayhan
Team: Batuhan Yenice, Fatih Eke, Talha Güzel
Keywords: 3D Modeling, Stereo Vision, Image Processing, Computer Vision
Abstract: Some robotic applications require a 3D model of the environment. These applications include aerial, underwater, outdoor or extraterrestrial missions. However, it also applies to many local scenarios such as 3D models, navigation tasks. This study is limited to 3D mapping to support self-driving vehicles to be used in the city. In this study, the system will create real-time 3D maps for autonomous vehicles using the stereo camera and visual odometry method, and these maps are aimed to overlap with real-world metrics at a high rate, taking into account critical points on the road. These critical points are defined as the X, Y, Z coordinates of obstacles up to 1.5 meters high or their vectorial distances from a reference point in the vehicle. The stereo cameras were calibrated successfully and the distortions of the captured photos were successfully corrected. Thanks to the trained model, most objects on the road have been successfully identified. An outlier map was successfully created for each pair of right and left camera images, and the distance of each recognized object in the environment was highly matched with the real world data.
THE PROSTHETIC ROBOTIC HAND
Advisor: Asst. Prof.
Team: Eylül Coşar, İsmail Cem Tüzün
Keywords: EMG , Artificial Neural Network, Machine Learning, Gesture classification, MRMR
Abstract: This senior design project focuses on developing artificial neural networks and EMG-based prosthetic robotic hands to enhance mobility for individuals with upper limb loss. The project aims to provide a cost-effective solution for hand gesture mobility and has potential applications in industry standards. By collecting EMG data, performing gesture classification, and implementing machine learning algorithms, a functional prosthetic robotic hand system capable of recognizing and interpreting hand gestures is designed. The project considers public health, safety, and welfare requirements and aims to improve the quality of life for individuals with upper limb loss while also having social and economic impacts. Theoretical background on EMG and neural networks is provided, along with the use of the Maximum Relevance - Minimum Redundancy (MRMR) algorithm for feature selection. Several solution methods for EMG signal classification are reviewed, showcasing the potential of artificial neural networks in achieving high accuracy in recognizing hand gestures. Overall, this project contributes to the advancement of prosthetic technology and offers a promising solution for improving mobility and integration for individuals with limb loss.
DETERMINATION ELECTROMAGNETIC PARAMETERS USING WAVEGUIDE
Advisor: Asst. Prof.
Team: Yuşa Yusufoğlu, İrem Kuğu, Ahmet Aksoy
Keywords: Epsilon, Waveguide, NRW Algorithm, CST application, MATLAB
Abstract: The accurate determination of electromagnetic parameters plays a crucial role in various applications ranging from telecommunications to radar systems. This senior design project focuses on developing a methodology to determine electromagnetic parameters using a waveguide-based approach. By leveraging the principles of wave propagation and electromagnetic field analysis, the project aims to provide a reliable and efficient method for characterizing the electromagnetic properties of materials and devices. This senior design project focuses on the determination of electromagnetic parameters using waveguide simulation and MATLAB implementation. The project utilizes the CST application to simulate the waveguide and obtain S-parameters data. The acquired data is then processed and analyzed using MATLAB, which incorporates the NRW algorithm to detect epsilon values by observing the generated graphs. The project demonstrates the feasibility of this approach for accurately characterizing electromagnetic properties. The methodology offers a cost-effective and efficient solution for researchers and engineers in various fields such as telecommunications and microwave engineering.
ENHANCING CNN-BASED EMOTION RECOGNITION WITH DATA AUGMENTATION AND PREPROCESSING TECHNIQUES
Advisor: Asst. Prof.
Team: Bora Kayaoglu, Tolga Toktas
Keywords: Convolutional Neural Networks (CNN), Deep Learning, Emotion Recognition, Imbalanced Database, Image Processing, Data Augmentation, Synthetic Image Generation
Abstract: Emotion recognition is a process with the ability to understand people's emotional states and expressions. This ability is used in many different fields today. Understanding people's emotional states is essential in areas such as human-machine interaction, marketing and security. Understanding and recognizing emotional states allows to better meet people's needs and expectations and achieve more effective results. In this project, a system that recognizes emotion from human faces is designed using Convolutional Neural Networks (CNN), one of the Deep Learning algorithms. CNN performs moderately to good when trained with a database. However, the lack of accessible, large and balanced databases for the use of deep learning methods for emotion recognition prevents high performance. In recent studies, a success of 65 ± 5% has been achieved as a result of training the CNN model with the FER2013 database, and it is aimed to increase this performance by using various methods in this project. In order to overcome the imbalance in the databases and thus increase the system success, studies were carried out on the FER2013, FER+, CK+ and KDEF databases. The number of data is increased by combining various databases; The images in the databases are subjected to various preprocessing to reduce their differences before training the neural network. These pre-processes help the neural network work better by making facial images more homogeneous and standardized. Data augmentation and synthetic rendering methods are used to reduce the imbalance in the data distribution that remains in existence despite the increasing number of data in the consolidated database and to increase the performance. With the CNN-based method developed by using database merging, image preprocessing and data augmentation, emotion recognition can be achieved with 80.44% success.
AN INVESTIGATION OF QUADROTOR UAV SYSTEMS
Advisor: Asst. Prof.
Team: İrfan Akyavaş, Emir Hüseyin Çetin, Mustafa Ali İnan
Keywords: SOC, Quadrotor, UAV, Propeller, Thrust, Dynamical Model, KiBaM, Available/Unavailable Charge, Recovery Effect, Rate Capacity Effect
Abstract: Quadrotor UAV usage have been gaining traction int the last few years. With increased areas of application, need for differently configured UAVs arise. In order to meet these new specifications, different components may be used in the UAV. Our project aims to build a simulation environment for quadrotor UAVs in which user can use parameters from previous experiments/simulations for every subsystem of the UAV to simulate the UAV. If done accurately this would allow decreased design and prototyping costs since effects of different component combinations can be observed without the need of building a physical prototype.