Research Outputs
Permanent URI for this communityhttps://hdl.handle.net/20.500.14288/2
Browse
11 results
Search Results
Publication Open Access A gated fusion network for dynamic saliency prediction(Institute of Electrical and Electronics Engineers (IEEE), 2022) Kocak, Aysun; Erdem, Erkut; Department of Computer Engineering; Erdem, Aykut; Faculty Member; Department of Computer Engineering; College of Engineering; 20331Predicting saliency in videos is a challenging problem due to complex modeling of interactions between spatial and temporal information, especially when ever-changing, dynamic nature of videos is considered. Recently, researchers have proposed large-scale data sets and models that take advantage of deep learning as a way to understand what is important for video saliency. These approaches, however, learn to combine spatial and temporal features in a static manner and do not adapt themselves much to the changes in the video content. In this article, we introduce the gated fusion network for dynamic saliency (GFSalNet), the first deep saliency model capable of making predictions in a dynamic way via the gated fusion mechanism. Moreover, our model also exploits spatial and channelwise attention within a multiscale architecture that further allows for highly accurate predictions. We evaluate the proposed approach on a number of data sets, and our experimental analysis demonstrates that it outperforms or is highly competitive with the state of the art. Importantly, we show that it has a good generalization ability, and moreover, exploits temporal information more effectively via its adaptive fusion scheme.
Publication Metadata only A realistic simulation environment for mri-based robust control of untethered magnetic robots with intra-operational imaging(IEEE-Inst Electrical Electronics Engineers Inc, 2020) Tiryaki, Mehmet Efe; Erin, Önder; N/A; Department of Mechanical Engineering; Sitti, Metin; Faculty Member; Department of Mechanical Engineering; School of Medicine; College of Engineering; 297104Dual-use of magnetic resonance imaging (MRI) devices for robot tracking and actuation has transformed them into potential medical robotics platforms for targeted therapies and minimally invasive surgeries. In this letter, we present the dynamic simulations of anMRI-based tracking and actuation scheme, which performs intra-operational imaging while controlling untethered magnetic robots. In our realistic rigid-body simulation, we show that the robot could be controlled with a 1D projection-based position feedback while performing intra-operational echo-planar imaging (EPI). From the simulations, we observe that the velocity estimation error is the main source of the controller instability for low MRI sequence frequencies. To minimize the velocity estimation errors, we constrain the controller gains according to maximum closed-loop rates achievable for different sequence durations. Using the constrained controller in simulations, we confirm that EPI imaging could be introduced to the sequence as an intra-operational imaging method. Although the intro-operational imaging increases the position estimation error to 2.0 mm for a simulated MRI-based position sensing with a 0.6 mm Gaussian noise, it does not cause controller instability up to 128 k-space lines.With the presented approach, continuous physiological images could be acquired during medical operations while a magnetic robot is actuated and tracked inside an MRI device.Publication Metadata only An adaptive admittance controller for collaborative drilling with a robot based on subtask classification via deep learning(Elsevier, 2022) Aydin, Yusuf; N/A; N/A; N/A; Department of Mechanical Engineering; Güler, Berk; Niaz, Pouya Pourakbarian; Madani, Alireza; Başdoğan, Çağatay; Master Student; Master Student; Master Student; Faculty Member; Department of Mechanical Engineering; Graduate School of Sciences and Engineering; Graduate School of Sciences and Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; N/A; N/A; 125489In this paper, we propose a supervised learning approach based on an Artificial Neural Network (ANN) model for real-time classification of subtasks in a physical human-robot interaction (pHRI) task involving contact with a stiff environment. In this regard, we consider three subtasks for a given pHRI task: Idle, Driving, and Contact. Based on this classification, the parameters of an admittance controller that regulates the interaction between human and robot are adjusted adaptively in real time to make the robot more transparent to the operator (i.e. less resistant) during the Driving phase and more stable during the Contact phase. The Idle phase is primarily used to detect the initiation of task. Experimental results have shown that the ANN model can learn to detect the subtasks under different admittance controller conditions with an accuracy of 98% for 12 participants. Finally, we show that the admittance adaptation based on the proposed subtask classifier leads to 20% lower human effort (i.e. higher transparency) in the Driving phase and 25% lower oscillation amplitude (i.e. higher stability) during drilling in the Contact phase compared to an admittance controller with fixed parameters.Publication Open Access Children's reliance on the non-verbal cues of a robot versus a human(Public Library of Science, 2019) Verhagen J.; Van Den Berghe R.; Oudgenoeg-Paz O.; Leseman P.; Department of Psychology; Küntay, Aylin C.; Faculty Member; Department of Psychology; College of Social Sciences and Humanities; 178879Robots are used for language tutoring increasingly often, and commonly programmed to display non-verbal communicative cues such as eye gaze and pointing during robot-child interactions. With a human speaker, children rely more strongly on non-verbal cues (pointing) than on verbal cues (labeling) if these cues are in conflict. However, we do not know how children weigh the non-verbal cues of a robot. Here, we assessed whether four- to six-year-old children (i) differed in their weighing of non-verbal cues (pointing, eye gaze) and verbal cues provided by a robot versus a human; (ii) weighed non-verbal cues differently depending on whether these contrasted with a novel or familiar label; and (iii) relied differently on a robot's non-verbal cues depending on the degree to which they attributed human-like properties to the robot. The results showed that children generally followed pointing over labeling, in line with earlier research. Children did not rely more strongly on the non-verbal cues of a robot versus those of a human. Regarding pointing, children who perceived the robot as more human-like relied on pointing more strongly when it contrasted with a novel label versus a familiar label, but children who perceived the robot as less human-like did not show this difference. Regarding eye gaze, children relied more strongly on the gaze cue when it contrasted with a novel versus a familiar label, and no effect of anthropomorphism was found. Taken together, these results show no difference in the degree to which children rely on non-verbal cues of a robot versus those of a human and provide preliminary evidence that differences in anthropomorphism may interact with children's reliance on a robot's non-verbal behaviors.Publication Open Access Control and transport of passive particles using self-organized spinning micro-disks(Institute of Electrical and Electronics Engineers (IEEE), 2022) Basualdo, Franco N. Pinan; Gardi, Gaurav; Wang, Wendong; Demir, Sinan O.; Bolopion, Aude; Gauthier, Michael; Lambert, Pierre; Department of Mechanical Engineering; Sitti, Metin; Faculty Member; Department of Mechanical Engineering; College of Engineering; School of Medicine; 297104Traditional robotic systems have proven to be instrumental in object manipulation tasks for automated manufacturing processes. Object manipulation in such cases typically involves transport, pick-and-place and assembly of objects using automated conveyors and robotic arms. However, the forces at microscopic scales (e.g., surface tension, Van der Waals, electrostatic) can be qualitatively and quantitatively different from those at macroscopic scales. These forces render the release of objects difficult, and hence, traditional systems cannot be directly transferred to small scales (below a few millimeters). Consequently, novel micro-robotic manipulation systems have to be designed to take into account these scaling effects. Such systems could be beneficial for micro-fabrication processes and for biological studies. Here, we show autonomous position control of passive particles floating at the air-water interface using a collective of self-organized spinning micro-disks with a diameter of 300 mu m. First, we show that the spinning micro-disks collectives generate azimuthal flows that cause passive particles to orbit around them. We then develop a closed-loop controller to demonstrate autonomous position control of passive particles without physical contact. Finally, we showcase the capability of our system to split from an expanded to several circular collectives while holding the particle at a fixed target. Our system's contact-free object manipulation capability could be used for transporting delicate biological objects and for guiding self-assembly of passive objects for micro-fabrication.Publication Open Access Deep learning-based 3D magnetic microrobot tracking using 2D MR images(Institute of Electrical and Electronics Engineers (IEEE), 2022) Tiryaki, Mehmet Efe; Demir, Sinan Özgün; Department of Mechanical Engineering; Sitti, Metin; Faculty Member; Department of Mechanical Engineering; College of Engineering; School of Medicine; 297104Magnetic resonance imaging (MRI)-guided robots emerged as a promising tool for minimally invasive medical operations. Recently, MRI scanners have been proposed for actuating and localizing magnetic microrobots in the patient's body using two-dimensional (2D) MR images. However, three-dimensional (3D) magnetic microrobots tracking during motion is still an untackled issue in MRI-powered microrobotics. Here, we present a deep learning-based 3D magnetic microrobot tracking method using 2D MR images during microrobot motion. The proposed method comprises a convolutional neural network (CNN) and complementary particle filter for 3D microrobot tracking. The CNN localizes the microrobot position relative to the 2D MRI slice and classifies the microrobot visibility in the MR images. First, we create an ultrasound (US) imaging-mentored MRI-based microrobot imaging and actuation system to train the CNN. Then, we trained the CNN using the MRI data generated by automated experiments using US image-based visual servoing of a microrobot with a 500 mu m-diameter magnetic core. We showed that the proposed CNN can localize the microrobot and classified its visibility in an in vitro environment with +/- 0.56 mm and 87.5% accuracy in 2D MR images, respectively. Furthermore, we demonstrated ex-vivo 3D microrobot tracking with +/- 1.43 mm accuracy, improving tracking accuracy by 60% compared to the previous studies. The presented tracking strategy will enable MRI-powered microrobots to be used in high-precision targeted medical applications in the future.Publication Metadata only Developing a prototyping method for involving children in the design of classroom robots(2018) Obaid, Mohammad; Barendregt, Wolmet; N/A; Department of Media and Visual Arts; Baykal, Gökçe Elif; Yantaç, Asım Evren; Researcher; Faculty Member; Department of Media and Visual Arts; KU Arçelik Research Center for Creative Industries (KUAR) / KU Arçelik Yaratıcı Endüstriler Uygulama ve Araştırma Merkezi (KUAR); Graduate School of Social Sciences and Humanities; College of Social Sciences and Humanities; N/A; 52621Including children in the design of technologies that will have an impact on their daily lives is one of the pillars of user-centered design. Educational robots are an example of such a technology where children’s involvement is important. However, the form in which this involvement should take place is still unclear. Children do not have a lot of experience with educational robots yet, while they do have some ideas of what robot could be like from popular media, such as BayMax from the Big Hero 6 movie. In this paper we describe two pilot studies to inform the development of an elicitation method focusing on form factors; a first study in which we have asked children between 8 and 15 years old to design their own classroom robot using a toolkit, the Robo2Box, and a second study where we have compared the use of the Robo2Box toolkit and clay as elicitation methods. We present the results of the two studies, and discuss the implications of the outcomes to inform further development of the Robo2Box for prototyping classroom robots by childrenPublication Metadata only Effect of Al2O3 and ZrO2 filler material on the microstructural, thermal and dielectric properties of borosilicate glass-ceramics(Multidisciplinary Digital Publishing Institute (MDPI), 2023) Karaahmet, Oğuz; Çiçek, Buğra; N/A; Department of Chemistry; Arıbuğa, Dilara; Balcı, Özge; PhD Student; Researcher; Department of Chemistry; Graduate School of Sciences and Engineering; College of Sciences; N/A; 295531Various glass-ceramics are widely used or considered for use as components of microelectronic materials due to their promising properties. In this study, borosilicate glass was prepared using the powder metallurgical route and then mixed with different amounts of Al2O3 and ZrO2 filler materials. Glass-ceramics are produced by high-energy ball milling and conventional sintering process under Ar or air. In this study, the effects of different filler materials and different atmospheres on the microstructural, thermal and dielectric properties were investigated. The data showed that ZrO2 filler material led to better results than Al2O3 under identical working conditions and similar composite structures. ZrO2 filler material significantly enhanced the densification process of glass-ceramics (100% relative density) and led to a thermal conductivity of 2.904 W/K.m, a dielectric constant of 3.97 (at 5 MHz) and a dielectric loss of 0.0340 (at 5 MHz) for the glass with 30 wt.% ZrO2 sample. This paper suggests that prepared borosilicate glass-ceramics have strong sinterability, high thermal conductivity, and low dielectric constants, making them promising candidates for microelectronic devices.Publication Metadata only Guidelines for designing social robots as second language tutors(Springer, 2018) Belpaeme, Tony; Vogt, Paul; van den Berghe, Rianne; Bergmann, Kirsten; de Haas, Mirjam; Kennedy, James; Oudgenoeg-Paz, Ora; Papadopoulos, Fotios; Schodde, Thorsten; Verhagen, Josje; Wallbridge, Christopher D.; Willemsen, Bram; de Wit, Jan; Hoffmann, Laura; Kopp, Stefan; Krahmer, Emiel; Montanier, Jean-Marc; Pandey, Amit Kumar; Department of Psychology; Department of Psychology; Department of Psychology; Department of Psychology; Department of Psychology; Department of Psychology; Göksun, Tilbe; Kanero, Junko; Küntay, Aylin C.; Geçkin, Vasfiye; Mamuş, Ayşe Ezgi; Oranç, Cansu; Faculty Member; Researcher; Faculty Member; Researcher; Researcher; Researcher; Department of Psychology; College of Social Sciences and Humanities; College of Social Sciences and Humanities; College of Social Sciences and Humanities; College of Social Sciences and Humanities; College of Social Sciences and Humanities; College of Social Sciences and Humanities; 47278; N/A; 178879; N/A; N/A; N/AIn recent years, it has been suggested that social robots have potential as tutors and educators for both children and adults. While robots have been shown to be effective in teaching knowledge and skill-based topics, we wish to explore how social robots can be used to tutor a second language to young children. As language learning relies on situated, grounded and social learning, in which interaction and repeated practice are central, social robots hold promise as educational tools for supporting second language learning. This paper surveys the developmental psychology of second language learning and suggests an agenda to study how core concepts of second language learning can be taught by a social robot. It suggests guidelines for designing robot tutors based on observations of second language learning in human-human scenarios, various technical aspects and early studies regarding the effectiveness of social robots as second language tutors.Publication Metadata only Repetitive control of an XYZ piezo-stage for faster nano-scanning: numerical simulations and experiments(Pergamon-Elsevier Science Ltd, 2011) Necipoğlu, Serkan; Güvenç, Levent; N/A; Department of Mechanical Engineering; N/A; Cebeci, Selman; Başdoğan, Çağatay; Has, Yunus Emre; Master Student; Faculty Member; Master Student; Department of Mechanical Engineering; Graduate School of Sciences and Engineering; College of Engineering; Graduate School of Sciences and Engineering; N/A; 125489; N/AA repetitive controller (RC) is implemented to control the Z-axis movements of a piezo-scanner used for AFM scanning and then tested through scan experiments and numerical simulations. The experimental and simulation results show that the RC compensates phase delays better than the standard PI controller at high scan speeds, which leads to less scan error and lower interaction forces between the scanning probe and the surface being scanned. Since the AFM experiments are not perfectly repeatable in the physical world, the optimum phase compensators of the RC resulting this performance are determined through the numerical simulations performed in MATLAB/Simulink. Furthermore, the numerical simulations are also performed to show that the proposed RC is robust and does not require re-tuning of these compensators when the consecutive scan lines are not similar and a change occurs in the probe characteristics. (C) 2011 Elsevier Ltd. All rights reserved.