|
|
Human-like conversations with Digital Personas: Conversational strategies that support mutual understanding
Kunc, Benjamín ; Šipošová, Barbora (advisor) ; Nováková Schöffelová, Miroslava (referee)
Communication is a collaborative process, and as such requires the communicators to create common ground - the assumption that they mutually understand each other on a sufficient level (Clark, 1996). The AI-powered virtual agents capable of using natural language to communicate with humans have recently attracted the attention of both the general public and experts, as the progress enables humans to give virtual agents more complex tasks in which a conversation itself is a crucial part of a task, such as in education, healthcare or mental health. While the virtual agents are getting better at understanding natural language, their ability to fulfill complex independent tasks (conducting semi-structured interviews, tutoring, coaching, etc.) is bounded by the limitations of their communication skills, and thus they perform the best at rather short, domain-specific conversations (Drouin et al., 2022). The goals of this thesis were to 1) create an experimental framework designed for a brief (5min) social voice chat between humans and virtual agents that would allow to experimentally manipulate different conversational strategies; 2) propose the first set of conversational strategies for mutual understanding and suggest appropriate self-report and behavioral metrics to measure the impact of the...
|
|
|
Three virtual characters in a quarrel: an interactive model
Babor, Petr ; Brom, Cyril (advisor) ; Beneš, Jan (referee)
In computer games, with the ambition to tell a story are often cut-scenes, passages of games, where the player does not play, but only follows a part of the narrated story. This fact distract player from playing. These passages should be integrated as much as possible in the game itself. The goal of our work is to design, implement and test a model for creating cut-scenes with dynamic development and limited user interaction possibilities. We restricted the model to a situation where three characters arguing and we implemented it for the Pogamut platform. We use animations and emoticons as means of expression, characters do not speak by natural language. Movement is ensured by the steering rules from the project Steering Library. Dramaturgy and synchronization of scenes is controlled by a stochastic finite state machine. We made an experiment where the test subjects watched the generated cut-scenes and compared them with hand-made to evaluate the work. The results confirmed the assumption that the quality of generated scenes do not reach handwritten ones. However the test subjects still perceived them positively, because of the ability to influence their development. Implemented parts of the model can be used to create games and educational programs with social theme. Theoretical design of the model can be...
|
|
|
Adaptive formations for virtual agents
Švagerka, Michal ; Plch, Tomáš (advisor) ; Čermák, Miroslav (referee)
There is a growing number of virtual agents in today's virtual worlds. This is directly related to the need of coordinating their behavior and movement. In this work we design algorithms to maintain desired formation of agents while moving through difficult terrain. The methods we study should maintain the requirements of the formation as well as traverse along various obstacles. We then use an arbitrary fitness function to compare the performance of these methods on typical scenarios.
|
|
|
Episodická pamäť virtuálnych agentov: Klamné spomienky
Čermák, Michal ; Brom, Cyril (advisor) ; Pešková, Klára (referee)
The goal of this work is to design a model of episodic memory for virtual agents capable of creating false memories and implement its prototype. The model architecture is inspired by present day knowledge about human episodic memory. Core structures in our model are the chronobags storing details of experienced episodes, and the schemabag storing the general scripts for all experienced events. The episodes are stored in the form of AND-OR trees and it is possible to derive missing details from the stored scripts during recall. We followed the work of Burkert and implemented a prototype of designed memory model in Java using the Pogamut 3 Framework. The model is parameterizable and it is evaluated in several experiments. In these experiments we looked on the trends in correct and incorrect recalls over time and we simulated the experiment demonstrating existence of room schemata. We compared the results produced by our model with the data from psychological researches and we showed that our model can produce false memories similar to false memories recollected by humans.
|
|
|
Coordination of multiple virtual agents in team-based computer games
Kolombo, Martin ; Gemrot, Jakub (advisor) ; Kadlec, Rudolf (referee)
Práce se zabývá problémem efektivní koordinace týmu virtuálních agent· v počí- tačových hrách. Hlavním cílem bylo navrhnou obecnou architektru pro práci s informacemi o virtuálním prostředí, která poskytuje programátorovi dobrý přístup k týmovým znalostem o virtuálním prostředí. Zvolená architektura neklade na programátora žádné požadavky ohledně implementace týmu agent· a nezávisí na žádném konkrétním virtuálním prostředí. Architektura je založena na přirozeném rozdělení dat podle jejich schopnosti se měnit v pr·běhu simulace a podle jejich subjektivity. Tato distribuce dat přirozeně vytváří sdílené znalosti týmu a programátorovi tak umožní založit rozhodování agent· na znalostech celého týmu namísto pouze jediného agenta. Architekturu jsme implemento- vali do systému Pogamut, ve kterém jsme provedli validaci zpětné kompati- bility. Vzheldem k nekompatibilitě formátu dat exportovaných součástí Poga- mutu, který jsme objevili v poslední fázi validace jsme provedli validaci pouze částečnou. Úpravu platformy Pogamut a následnou implementaci týmu agent· využívajícího naši architekturu plánujeme jako pokračování práce. 1
|
|
|
StarCraft and Emergent in Pogamut 3 environment
Dekar, Martin ; Brom, Cyril (advisor) ; Děchtěrenko, Filip (referee)
The Pogamut toolkit designed for rapid prototyping of computer game agents has been so far used for prototyping the agents based on 3D FPS Unreal Tournament 2004 and its sequels. After the environment of RTS Defcon was connected to Pogamut a question arose how difficult it would be to connect some other significantly different environments and action selection mechanisms. In order to test this flexibility of Pogamut we have interconnected it with more complex RTS video game StarCraft:Brood War and large neural network simulator Emergent, together with Jason and POSH action selection mechanisms. The work analyzes created connections to detail and demonstrates their functionality on examples. An integral part of the work is also web with video tutorials and guides. In this work we also analyze Pogamut's readiness to be connected to other environments.
|
|
|
Episodická pamäť virtuálnych agentov: Klamné spomienky
Čermák, Michal ; Brom, Cyril (advisor) ; Pešková, Klára (referee)
The goal of this work is to design a model of episodic memory for virtual agents capable of creating false memories and implement its prototype. The model architecture is inspired by present day knowledge about human episodic memory. Core structures in our model are the chronobags storing details of experienced episodes, and the schemabag storing the general scripts for all experienced events. The episodes are stored in the form of AND-OR trees and it is possible to derive missing details from the stored scripts during recall. We followed the work of Burkert and implemented a prototype of designed memory model in Java using the Pogamut 3 Framework. The model is parameterizable and it is evaluated in several experiments. In these experiments we looked on the trends in correct and incorrect recalls over time and we simulated the experiment demonstrating existence of room schemata. We compared the results produced by our model with the data from psychological researches and we showed that our model can produce false memories similar to false memories recollected by humans.
|
|
|
Level-of detail AI: a prototype of a large virtual world
Skupien, David ; Brom, Cyril (advisor) ; Vorba, Jiří (referee)
Title: Level of Detail AI: a prototype of a large virtual world Author: David Skupien Department: Department of Software and Computer Science Education Supervisor: Mgr. Cyril Brom, Ph.D. Abstract: The Level Of Detail technique (LOD) is an established means of optimization in computer graphics. It uses the limitations of human perception to spare system resources. Progressively, this approach has found its use in artificial intelligence. This adaptation, commonly known as LOD AI, is used to optimize autonomous agent control in large virval worlds. This thesis consists of two parts. The first part deals with standing LOD AI uses, mostly focusing on the IVE project. The second part directly uses the IVE project to implement a LOD AI game using the LOD based reactive planning algorithm called S-GHRP over the ISMA representation. The above game is then tested on respondents to verify the practical usability of S-GHRP/ISMA. In its conclusion, this thesis briefly ponders the possibilities of further development of techniques used in the IVE project, based on previous findings. Keywords: artificial intelligence, LOD AI, virtual agent, large virtual world, IVE
|
|
|
Coordination of multiple virtual agents in team-based computer games
Kolombo, Martin ; Gemrot, Jakub (advisor) ; Kadlec, Rudolf (referee)
Práce se zabývá problémem efektivní koordinace týmu virtuálních agent· v počí- tačových hrách. Hlavním cílem bylo navrhnou obecnou architektru pro práci s informacemi o virtuálním prostředí, která poskytuje programátorovi dobrý přístup k týmovým znalostem o virtuálním prostředí. Zvolená architektura neklade na programátora žádné požadavky ohledně implementace týmu agent· a nezávisí na žádném konkrétním virtuálním prostředí. Architektura je založena na přirozeném rozdělení dat podle jejich schopnosti se měnit v pr·běhu simulace a podle jejich subjektivity. Tato distribuce dat přirozeně vytváří sdílené znalosti týmu a programátorovi tak umožní založit rozhodování agent· na znalostech celého týmu namísto pouze jediného agenta. Architekturu jsme implemento- vali do systému Pogamut, ve kterém jsme provedli validaci zpětné kompati- bility. Vzheldem k nekompatibilitě formátu dat exportovaných součástí Poga- mutu, který jsme objevili v poslední fázi validace jsme provedli validaci pouze částečnou. Úpravu platformy Pogamut a následnou implementaci týmu agent· využívajícího naši architekturu plánujeme jako pokračování práce. 1
|
|
|
Three virtual characters in a quarrel: an interactive model
Babor, Petr ; Brom, Cyril (advisor) ; Beneš, Jan (referee)
In computer games, with the ambition to tell a story are often cut-scenes, passages of games, where the player does not play, but only follows a part of the narrated story. This fact distract player from playing. These passages should be integrated as much as possible in the game itself. The goal of our work is to design, implement and test a model for creating cut-scenes with dynamic development and limited user interaction possibilities. We restricted the model to a situation where three characters arguing and we implemented it for the Pogamut platform. We use animations and emoticons as means of expression, characters do not speak by natural language. Movement is ensured by the steering rules from the project Steering Library. Dramaturgy and synchronization of scenes is controlled by a stochastic finite state machine. We made an experiment where the test subjects watched the generated cut-scenes and compared them with hand-made to evaluate the work. The results confirmed the assumption that the quality of generated scenes do not reach handwritten ones. However the test subjects still perceived them positively, because of the ability to influence their development. Implemented parts of the model can be used to create games and educational programs with social theme. Theoretical design of the model can be...
|
guest ::
