Dr. Hakki Erhan Sevil

This paper presents an autonomous multi-agent unmanned aerial vehicle (UAV) system designed to perform object detection through Visual Question Answering (VQA) using aerial imagery. The system utilizes an entropy-based distributed behavior model to coordinate UAV movements toward designated
waypoints. A VQA model is used to analyze aerial footage for detection of objects of interest. The study investigates the impact of various distributed behavior configurations, including number of UAVs, UAV formations, flight altitude, and separation distance. After analysis, a final optimized configuration for maximizing surface area coverage and VQA model performance were found. These findings contribute to the development of aerial systems capable of collaborative visual reasoning in complex environments.

Image segmentation is essential for navigation and scene understanding in autonomous systems, particularly in unstructured outdoor environments. This study investigates the segmentation capabilities of DALL-E 3, a generative text-to-image model, that is not explicitly trained for semantic segmentation.
A custom segmentation pipeline was developed to evaluate and refine DALL-E 3 outputs on outdoor images from the RELLIS-3D dataset. The post-processing workflow includes morphological operations with varied structure elements to enhance segmentation accuracy. Segmentation accuracy was assessed using mean Intersection over Union (mIoU) across selected terrain classes. Results show that the raw DALL-E 3 outputs were improved after
developed post-processing refinement, and resulting accuracy values are competitive with supervised models, HRNet+OCR and GSCNN. These results demonstrate that text-to-image models, when paired with domain-aware post-processing, offer a promis-ing alternative for flexible, rapid-deployment segmentation for universal robotics without requiring labeled training data. These efforts contribute to our research team’s broader goal of enabling
intelligent mobile robots capable of autonomous perception and decision-making in complex environments.

In this study, we investigate the feasibility of detecting post-disaster damages through camera images obtained onboard an Unmanned Aerial Vehicle (UAV). Aerial footage from the University of West Florida (UWF) after being hit by hurricane Sally in 2020 is used in our study. Our goal is to automatically locate and identify all the roof damages caused by Sally on the university campus and compare two methods of detection. The first is a Convolutional Neural Network (CNN) based approach and the second is a cascade of classifiers model. We utilize
cascading classifiers from the OpenCV Python library and a TensorFlow Object Detection API model both retrained on images hand annotated by our team to demonstrate the damage detection capabilities of these models. The aim of this study is to analyze feasibility and compare results between CNN and cascade classifier model for post-disaster damage detection to aid the effort of damage recovery after hurricanes.

Empirical studies suggest that traffic flow generally exhibits irregular and complex behavior. Modeling of traffic flow characteristics is difficult and needs new techniques. In this study, we analyzed chaotic structure in traffic time series data collected from an urban arterial in Istanbul over a period of about 1 week. Nonlinear techniques (correlation dimension and metric entropy) are used to identify chaotic structure. After detecting chaotic characteristics, the predictability of time series data was examined. It is found that the traffic flow at the main road of lkitelli - Mahmutbey location displayed a periodicity close to 24 hrs, and a 100 minute long prediction interval which is indicative of low dimensional chaos as found from the computed metric entropy. Traffic time series data included speed, occupancy rates, and volume at each lane on the main road of lkitelli - Mahmutbey on the European side.

In this study, a series of condition monitoring experiments on drill bits were carried out. Small diameter drill bit run-to-failure test rig was constructed and the prediction tests were performed. In the experiments, 10 small drill bits ( 1 mm ⌀ ) were tested until they broke down, while vibration data were consecutively taken in equal time intervals. A consistent decrement in variation of metric entropy just before the breakage was observed. As a result of the experiment results, metric entropy variation could be implemented as an early warning system.