Research
The overarching goal of my research is to create autonomous, distinctive, and embodied virtual characters that behave like humans. Such characters have enormous potential to promote accessibility in areas dependent on social interactions such as education, healthcare, and assistive technologies to accommodate people with diverse needs, backgrounds, and interests.
A Conversational Agent Framework with Multimodal personality expression
Consistently exhibited personalities are crucial elements of realistic, engaging, and behavior-rich conversational virtual agents. Both nonverbal and verbal cues help convey these agents’ unseen psychological states, contributing to our effective communication with them. We introduce a comprehensive framework to design conversational agents that express OCEAN personality factors through non-verbal behaviors like body movement and facial expressions, as well as verbal behaviors like dialogue selection and voice transformation. The framework combines existing personality expression methods with novel ones such as new algorithms to convey Laban Shape and Effort qualities. We perform Amazon Mechanical Turk studies to analyze how different communication modalities influence our perception of virtual agent personalities and compare their individual and combined effects on each personality dimension. The results indicate that our personality-based modifications are perceived as natural, and each additional modality improves perception accuracy, with the best performance achieved when all the modalities are present.
Publications
Sinan Sonlu, Uğur Güdükbay, and Funda Durupinar. 2021. A Conversational Agent Framework with Multi-modal Personality Expression. ACM Trans. Graph. 40, 1, Article 7 (January 2021). DOI:https://doi.org/10.1145/3439795. Download: [pdf]
Personality-Driven GAZE ANIMATION
We present a conditional generative adversarial learning approach to synthesize the gaze behavior of a given personality. Training is done using an existing data set that comprises eye-tracking data and personality traits of 42 participants performing an everyday task. Given the values of Big-Five personality traits (openness, conscientiousness, extroversion, agreeableness, and neuroticism), our model generates time series data consisting of gaze target, blinking times, and pupil dimensions. We use the generated data to synthesize the gaze motion of virtual agents on a game engine.
Publications
Funda Durupinar, Personality-Driven Gaze Animation with Conditional Generative Adversarial Networks. The 13th Annual ACM SIGGRAPH Conference on Motion, Interaction and Games (MIG'20), 2020. Download [pdf]
Funda Durupinar, Personality-Driven Gaze Animation with Conditional Generative Adversarial Networks, 2020. https://arxiv.org/abs/2012.02224
Personality-Driven Character Motion
A major goal of research on virtual humans is the animation of expressive characters that display distinct psychological attributes. Personality is a major factor in our assessment of other individuals due to its stable and definitive qualities. Humans convey a lot of information about their personalities through their body movements. Based on these facts, we describe a formal, broadly applicable, procedural, and empirically grounded association between personality and body motion and apply this association to modify any given virtual human body animation. We developed an expressive human motion generation system with the help of movement experts and conducted a crowdsourcing user study to explore how personality was perceived in motions with different styles. We then applied the findings to procedurally animate expressive characters with personality, and validated the generalizability of our approach across different models and animations via another perception study.
Publications
Funda Durupinar, Mubbasir Kapadia, Susan Deutsch, Michael Neff and Norman I. Badler. PERFORM: Perceptual Approach for Adding OCEAN Personality to Human Motion using Laban Movement Analysis. ACM Transactions on Graphics, 36(1), October 2016.
Funda Durupinar, Kuan Wang, Ani Nenkova and Norman I. Badler. An Environment for Transforming Game Character Animations Based on Nationality and Profession Personality Stereotypes. The Twelfth AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE-16), 2016. Download: [pdf]
Character Animation Based on Nationality and Profession Stereotypes
The animation of game characters is a key application area for creating realistic-looking human-like motion. Modeling a viewer’s expectations about the personality of commonly encountered groups allows us to provide developers with reasonable starting points for animation of a large set of possible game settings. The developers can use the stereotype, deliberately manipulate the stereotype to create a desired effect on the perception of the player (for example by creating characters that challenge the stereotype) or exaggerate a stereotype to the point of a caricature which may help players reconsider their own bias.
We developed a flexible environment to endow game characters with personality and apply these to the perception of nationality and profession stereotypes. We performed user studies on Amazon Mechanical Turk to find stereotypical associations between personality traits and 135 nationalities and 100 professions. We then evaluated the perception of these associations on animated humanoid characters. The studies show that there are indeed stereotypes associated with many nationalities and professions and that they can be conveyed through motion styles. Application areas extend beyond game characters to embodied agents, personal avatars, or motion picture pre-visualization.
Publications
Funda Durupinar, Kuan Wang, Ani Nenkova and Norman I. Badler. An Environment for Transforming Game Character Animations Based on Nationality and Profession Personality Stereotypes. The Twelfth AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE-16), 2016. Download: [pdf]
Oshin Agarwal, Funda Durupinar, Norman I. Badler and Ani Nenkova Word Embeddings (Also) Encode Human Personality Stereotype, *SEM 2019. https://www.aclweb.org/anthology/S19-1023
Psychological ParameterS for Crowd Simulation
In the social psychology literature, crowds are classified as audiences and mobs. Audiences are passive crowds, whereas mobs are active crowds with emotional, irrational and seemingly homogeneous behavior. In this study, we aim to create a system that enables the specification of different crowd types ranging from audiences to mobs. In order to achieve this goal we parametrize the common properties of mobs to create collective misbehavior. Because mobs are characterized by emotionality, we describe a framework that associates psychological components with individual agents comprising a crowd and yields emergent behaviors in the crowd as a whole. To explore the effectiveness of our framework we demonstrate two scenarios simulating the behavior of distinct mob types. In order to tune and test emotion contagion and personality parameters, we introduce a data-driven evaluation platform to compare our scenarios with videos of real-life events.
Publications
A. E. Basak, U. Gudukbay, F. Durupinar. Using Real Life Incidents for Realistic Virtual Crowds with Data-Driven Emotion Contagion, Computers and Graphics, 2018. Download: [pdf]
Ahmet Eren Basak, Funda Durupinar, Ugur Gudukbay, Learning From Real-Life Experiences: A Data-Driven Emotion Contagion Approach Towards More Realistic Virtual Crowds. The 30th International Conference on Computer Animation and Social Agents (CASA 2017).Download: [pdf]
Funda Durupinar, Ugur Gudukbay, Aytek Aman and Norman I. Badler. Psychological Parameters for Crowd Simulation: from Audiences to Mobs. IEEE Transactions on Visualization and Computer Graphics , Vol. 22, No. 9, pp. 2145-2159, September 2016. Download: [pdf]
Work presented at The ACM SIGGRAPH / Eurographics Symposium on Computer Animation (SCA'16), Zurich, Switzerland, 11-13 July 2016. Download: [slides]
Funda Durupinar, Ugur Gudukbay, Aytek Aman and Norman I. Badler. Simulation of Collective Crowd Behavior with Psychological Parameters in Simulating Heterogeneous Crowds with Interactive Behaviors, A K Peters/CRC Press, Nuria Pelechano, Jan M. Allbeck, Mubbasir Kapadia, Norman I. Badler(Eds), Nov 15 2016. ISBN: 9781498730365
Creating Crowd Variation with the OCEAN Personality Model
Personality is the sum of a person’s behavioral, temperamental, emotional, and mental traits. A popular model that describes personality is the Five Factor, or OCEAN (openness, conscientiousness, extroversion, agreeableness, and neuroticism) model. The personality space is composed of these five orthogonal dimensions.Each factor is bipolar and composed of several traits, which are essentially the adjectives that are used to describe people. We have mapped these trait terms to the low-level behavior parameters in the HiDAC (High-Density Autonomous Crowds) crowd simulation system.By incorporating a standard personality model to a high-density crowd simulation, our approach creates plausible variations in the crowd and enables novice users to dictate these variations.In order to verify the plausibility of our mapping we have conducted tests that evaluate users’ perception of the personality traits in the generated animations. We created several animations to examine how modifying the personality parameters of subgroups affects global crowd behavior. The animations exhibit the emergent behaviors of agents in scenarios in which the settings assigned according to the OCEAN model drive crowds behavior. In order to validate our system, we determined the correspondence between our mapping and the users’ perception of these trait terms in the videos. The results indicate a high correlation between our parameters and the participants perception of them.
Publications
Funda Durupınar, Nuria Pelechano, Jan M. Allbeck, Uğur Güdükbay, and Norman I. Badler. How the Ocean Personality Model Affects the Perception of Crowds. IEEE Computer Graphics and Applications, 31(3):22–31, 2011. Download: [pdf]
Funda Durupinar, Jan Allbeck, Nuria Pelechano, and Norman I. Badler. Creating Crowd Variation with the OCEAN Personality Model. Proceedings of Autonomous Agents and Multi-Agents Systems 2008. Download: [pdf]
Visualization of Crowd Synchronization on Footbridges
We propose a framework for the visualization of crowd walking synchronization on footbridges. The bridge is modeled as a mass-spring system, which is a weakly damped and driven harmonic oscillator. Both the bridge and the pedestrians walking on the bridge are affected by the movement of each other. The crowd acts according to local behavioral rules. Each pedestrian is provided with a kinematic walking system. We extend a current mathematical model of crowd synchronization on footbridges to include human walking model and crowd simulation techniques. We run experiments to evaluate the influence of these extensions on synchronization.
Publications
Funda Durupınar and Uğur Güdükbay. Visualization of Crowd Synchronization on Footbridges. Journal of Visualization, 13(1):69–77, 2010. Download: [pdf]
Dynamic Point-Region Quadtrees for Particle Simulations
We propose an algorithm for dynamically updating point-region (PR) quadtrees. Our algorithm is optimized for simultaneous update of data points comprising a quadtree. The intended application area focuses on simulating continuum phenomena, such as crowds, fluids, and smoke. We minimize the number of tree updates by making use of small changes in the positions of data points. We compare the efficiency of the proposed algorithm with two other approaches for updating a quadtree. One of these techniques creates the tree from scratch at each time-step. The second technique subsequently deletes a data point from the tree and reinserts it in its updated position. We achieve significant performance gains with our method in both cases.
Publications
Oğuzcan Oğuz, Funda Durupınar, and Uğur Güdükbay. Dynamic Point-region Quadtrees for Particle Simulations. Information Sciences, 218:133–145, 2013. Download: [pdf]
3D Garment Simulation and Rendering
We introduce a 3D graphics environment for virtual garment design and physically-based simulation. The system also provides rendering alternatives for the visualization of knitted and woven fabric, adopting a procedural approach that considers the physical structure of these fabric types.
Publications
Funda Durupınar and Uğur Güdükbay. Procedural Visualization of Knitwear and Woven Cloth. Computers & Graphics, 31(5):778–783, 2007. BibTeX Download: [pdf]
Funda Durupınar and Uğur Güdükbay. A Virtual Garment Design and Simulation System. In Proceedings of the 11th International Conference on Information Visualisation (IV '07), pp. 862–870, IEEE Computer Society, Zürich, Switzerland, July 2007. Download: [pdf]
Communicating with Computers
Detailed and comprehensive pathway information is a key element in understanding human genetic variation and disease mechanisms. Pathway information is currently in a fragmented state due to its rapid growth and insufficient efforts to assemble pieces of information into useful knowledge. We are developing a graphical biological pathway curation platform that focuses on curating detailed mechanistic pathways. The platform enables multiple humans, NLP and AI agents to collaborate in real-time on a common model using an event driven API. Our goal is to use this platform to explore disruptive technologies that can scale up biocuration such as NLP, human-computer collaboration, crowd-sourcing, alternative publishing and gamification. The platform includes a web-based graphical editor and an application server that provides model synchronization, connection to data sources and services, as well as novel interaction modes with human users and with computer actors. For the visualization and storage of data, we use community standards such as SBGN-PD and BioPAX.
Publications
Funda Durupinar-Babur, Metin Can Siper, Ugur Dogrusoz, Istemi Bahceci, Ozgun Babur and Emek Demir. Collaborative Workspaces for Pathway Curation. International Conference on Biological Ontology & BioCreative, 2016. Download: [pdf]