Under harsh geographical conditions where manned flight is not possible, the ability of the unmanned aerial vehicle (UAV) to successfully carry out the payload hold–release mission by avoiding obstacles depends on the optimal path planning and tracking performance of the UAV. The ability of the UAV to plan and track the path with minimum energy and time consumption is possible by using the flight parameters. This study performs the optimum path planning and tracking using Harris hawk optimization (HHO)–grey wolf optimization (GWO), a hybrid metaheuristic optimization algorithm, to enable the UAV to actualize the payload hold–release mission avoiding obstacles. In the study, the hybrid HHO–GWO algorithm, which stands out with its avoidance of local minima and speed convergence, is used to successfully obtain the feasible and effective path. In addition, the effect of the mass change uncertainty of the UAV on optimal path planning and tracking performance is determined. The effectiveness of the proposed approach is tested by comparing it with the metaheuristic swarm optimization algorithms such as particle swarm optimization (PSO) and GWO. The experimental results obtained indicate that the proposed algorithm generates a fast and safe optimal path without becoming stuck with local minima, and the quadcopter tracks the generated path with minimum energy and time consumption.

Modeling of unmanned aerial vehicle (UAV) with system identification is very important in terms of its model-based effective control. The modeling of UAV is required for aircraft crashes, analyzing autonomous aircrafts, preventing external disturbances, pre-flight analysis. However, since UAV has nonlinear inherent dynamics including inherent chaoticity and fractality, it becomes difficult to obtain a mathematical model under external disturbance. In this study, some of the inherent nonlinear dynamics of UAV are linearized and the model of UAV is obtained by system identification approaches under external disturbance. The linearized lateral dynamics of a fixed wing UAV is used in this study. Further, the flight motion equations applied to fixed wing UAV have been utilized for obtaining the coefficients of lateral model for straight and level flight. The roll angles are calculated using transfer functions for aileron, rudder and deflections inputs. The autoregressive exogenous (ARX), autoregressive moving average with exogenous (ARMAX) and output error (OE) parametric system identification approaches are performed to estimate UAV lateral dynamic system response as using empirical input-output data sets. The accuracy of parametric model estimation and model degrees are compared for different external disturbance effects.

Vehicle, industrial, and urban emissions remain major contributors to air quality degradation, affecting public health and the level of environmental cleanliness. Cost-effective specific pollutant estimation models, i.e., for carbon monoxide $\mathrm{CO}$, carbon dioxide ${\mathrm{CO}}_2$, and ammonia ${\mathrm{NH}}_3$, are essential to tackle the practical challenge of high-resolution monitoring for reducing vehicle emissions in traffic. Existing model design methods, however, may be insufficient, particularly for peak time estimations, since such models are typically designed using gridding-based vehicle-specific power polynomial and non-optimized artificial neural networks. In this paper, we propose vehicle emission models of pollutants based on a Bayesian Monte Carlo (MC) Dropout-based robust data-driven gated recurrent unit (BMC-GRU) method to enhance estimation robustness and mitigate the overfitting problem in the deep learning network. Bayesian optimization determines the optimal architecture by efficiently and probabilistically searching the hyperparameters of the network, while MC-Dropout quantifies epistemic uncertainty through multiple stochastic forward passes during testing. Therefore, the proposed method improves the models’ calibrations and robustness to distribution shifts. For benchmarking, least squares-based first- and fourth-order polynomials, conventional long-short term memory (LSTM), and bidirectional LSTM (BiLSTM)-based estimation models are designed. The proposed method outperforms the mentioned state-of-the-art methods with strong robust estimation performance. The experimental results on multiple real-world vehicle datasets demonstrate that the proposed method significantly outperforms state-of-the-art approaches. The method presents a promising solution for uncertainty-aware vehicle emission modeling that is applicable to transportation systems.

Otonom olarak hareket eden İnsansız Hava Araçlarında (İHA) yük taşıma sistemi sivil ve askeri alanlarda sıklıkla kullanılmaktadır. Coğrafi koşulların elverişsiz olduğu operasyon bölgelerinde belirlenen konumdan alınan faydalı yükün belirli bir rota takip edilerek hedeflenen konuma minimum hata ile bırakılması büyük önem arz etmektedir. Bu tür görevlerin başarıyla gerçekleştirilebilmesi için İHA’nın belirlenen rotayı minimum hata ile takip etmesi gerekmektedir. Bu çalışmada, yük alma bırakma sistemi kullanılarak belirlenen bölgedeki farklı renkli iki özdeş faydalı yükün dört pervaneli İHA tarafından sırasıyla alınarak belirlenen konuma bir rota üzerinden minimum hata ile bırakılabilmesi için bulanık PID denetleyicisi tasarlanmaktadır. Faydalı yük alma, taşıma ve bırakma görevi esnasında elde edilen sonuçlar PID denetleyici performansı ile kıyaslamalı olarak analiz edilmektedir.

Real-time prediction of the instantaneous fuel consumption rate (FCR) of any vehicle is the key to improving energy efficiency and reducing emissions. The conventional prediction methods, which include an on-board diagnostic (OBD) system, require the specific vehicle parameters and environmental conditions such as air density. We propose a data-driven Bayesian optimization and Monte Carlo (MC) Dropout methods-based long short-term memory (BMC-LSTM) network FCR prediction model using only the vehicle’s throttle position, velocity, and acceleration data. The cost-effective LSTM network-based solution enhances the high-resolution prediction accuracy within a deep learning framework. The network is integrated with the Bayesian optimization and MC-Dropout methods to ensure a probabilistically optimal hyperparameter set and robust networks. The proposed method presents an FCR model that provides calibrated predictions and reliability against distribution drift by probabilistically tuning hyperparameters with Bayesian optimization and quantifying epistemic uncertainty with the MC-Dropout. Our approach requires only vehicle speed, longitudinal acceleration, and throttle position at inference time. Note, however, that the reference FCR used to train and validate the models was obtained from OBD during data acquisition. The performance of the proposed method is compared with a conventional LSTM and Bidirectional LSTM-based multidimensional models, XGBoost and support vector regression-based models, and first- and fourth-order polynomials, which are derived using the least-squares method. The prediction performance of the method is evaluated using Mean Squared Error, Root Mean Squared Error, Mean Absolute, and R-squared statistical metrics. The proposed method achieves a superior $R^2$ score and substantially reduces the conventional error metrics.

Son yıllarda bilgisayar destekli tedavilerde yapay zekâ temelli uygulamalar diş hekimleri için hastalık teşhisinde kullanımı giderek artmaktadır. Diş hastalığın belirlenmesi sürecinde doğru bir şekilde bölütleme işlemi oldukça önemlidir. Diş bölütlenmesinin manuel olarak yapılması diş hekimleri tarafından yapılan teşhis için geçen süreyi ve işlem yükünü önemli ölçüde arttırmaktadır. Bu aşamada, makine öğrenmesi ve yapay zekâ yöntemleriyle otomatik olarak diş bölgesinin bölütlenmesi araştırmacıların ilgi duyduğu güncel bir konu olmaktadır. Çalışmada 12 diş stajyeri tarafından 15318 poligonlu 598 hastadan alınan X-Ray diş görüntüleri kullanılmaktadır. Kullanılan veri seti eğitim, doğrulama ve test olarak %70, %15, %15 olarak bölünmüştür. Bu veri seti otomatik olarak diş bölütlemeyi amaçlayan derin öğrenme ağının eğitim sürecinde kullanılmaktadır. SegFormer eğitim bloğu hiper parametrelerinin değişimine bağlı oluşturulan mimarilerinin performansları incelenmektedir. Burada MiT BO-B5 mimarilerine göre oluşturulan modellerin Dice benzerlik katsayılarına göre test verisi için performansları sırasıyla %92,61, %92,82, %93,25, %93,13, %93,17 ve %93,09 olarak elde edilmektedir. Elde edilen test sonuçlarına göre geliştirilen yapay zekâ tabanlı SegFormer ağı diş bölütlemeyi yüksek doğrulukla gerçekleştirmektedir. Geliştirilen derin öğrenme ağı özellikle diş hastalıklarının teşhisinde girdi olarak verimli bir şekilde kullanılabilecektir. Yüksek Dice benzerlik katsayıları, çalışmada sunulan SegFormer ağının diş bölgesini doğru bir şekilde tespit edebildiğini ifade etmektedir.