Abstract


Research behavior constructing methods in computer complexes


Introduction


The basic modern world wide web concept is the interactivity and quick response to events. Internet acquired a fundamentally new guise with the advent of Web 2.0 , as appeared possible a simple user without special knowledges and skills create his page without knowing the basic Internet principles.

High interactivity has elements of both the web space and offline applications today. These should include a variety of real action simulators used for training purposes, as well as games . Interactive computer systems includes the following systems:

  • online applications:
    • systems using advanced graphics technology(Flash, Silverlight etc)
    • multiplayer games (in general, using artificial intelligence)
  • offline applications:
    • Simulators
    • desktop games and gaming consoles (using modern devices, taking into account the player movements in the real world)


Relevance

Chosen topic relevance should be considered separately from the position of online and offline applications.
Interactive worlds in a browser prospects became very real with the announcement Adobe Flash Molehill tyechnology. Using GPU systems and 98% of plants copies of Adobe Flash around the world opens up the possibility for building complex systems that will be able to process over 3 million polygons without slowdowns. It is obvious that such systems will benefit everyone, regardless of operating system and browser.
As for offline applications, technologies Microsoft Kinect and PlayStation Move has a new features such as character interaction and orientation in space and taking into account user behavior. Microsoft Kinect technology is in the porting process to the Flash platform.
As for of behaviour developing, the concept of constructing scenarios for user interaction and the principles of artificial intelligence using various new technologies, developed early, require a substantial revision of the advent of advanced features.


Aims and objectives

The main purpose, which arise in the course of this study is to examine the basic concepts that have been established today in scope of behaviour scenarios costruction. We need to highlight areas that need to behaviour scenarios. The final step is to develop a new principle for script constructing that will take into account all the innovations that have appeared in recent years.


Technologies and System Overview


Microsoft Kinect и XNA (WPF)

Kinect is "controller without a controller" for Xbox 360. This technology includes:

  • Motion Sensor . Kinect uses a motion sensor that tracks every move. Therefore, you can manage not only the hands and palms. Involved the entire body. Hands, feet, knees, waist, hips and everything else
  • Track Skeleton . While driving, Kinect creates a digital representation of the skeleton, based on data depth. Therefore, if you move left or right or jumping, the sensor detects movement and introduces the user to the system
  • Face Detection . Kinect ID remembers the user, collecting physical data that is stored in the profile. Therefore, when starting a new game, Kinect learns the user allowed to join the game at any time

Originally Microsoft developed the device as opposed to PlayStation Move only for consoles. The device ported to PC today, enthusiastic was created a set of drivers. There is no official drivers for the PC from Microsoft yet.
Kinect offers great opportunities for emulate different kinds of behavior . With the official support for the PC advent, this device will serve as the basis for most simulators and trainers . With XNA Games support, games can be developed with the user-controller. It is possible to create applications with the interface, wich interaction is part of the body in the air using WPF.
Kinect is a horizontal box on a small round base, which is placed above or below the display. It's dimensions is approximately 23 inches long and 4 cm in height. It consists of two sensors, depth, color video camera and microphone array. Proprietary software carries a full 3-D recognition of body movements, facial expression and voice. Microphone array allows Xbox 360 to produce the sound source localization and noise suppression, which allows you to speak without a headphone and microphone Xbox Live. Depth sensor consists of infrared projector combined with a monochrome CMOS sensor that allows the sensor Kinect obtain three-dimensional image at any natural light.
Depth range and a program project allows you to automatically calibrate the sensor, taking into account playing conditions and environmental conditions, such as furniture in the room.
Judging by a recent patent Microsoft, Kinect be able to recognize sign language. While the patent applies only to ASL, but possibly other languages ​​will be added later. This is expected to reach a wider audience of users and will help to educate the dumb sign language. However, according to the official commentary, this feature is not included in the first version of Kinect because of lower costs in favor of resolution cameras. On the other hand, Microsoft does not renounce the use of the patent.


Adobe Flash and Molehill

In Feb, 2011 Adobe Corporation released Flash 11 incubator Build with new API Molehill. Molehill is the code name of the new method of displaying 2D and 3D graphics. This set of low-level commands for GPU , which includes advanced features between different screens and devices using the technology Adobe Flash . This new low-level interface provides enhanced 2D and 3D design applications, providing flexibility and movement of the load on the graphics core of the system, which significantly improves performance. Adobe Flash Player displays thousands of unsorted polygons around 30 Hz today. With the new software and hardware interface, a developer can use hundreds of thousands sorted polygons , which are displayed in high definition on the screen for 60 Hz. Using Molehill with Flash Player or Adobe Air you can create complex 3D world , which will be available on every computer or device connected to the Internet.

Most of the leading companies that was involved in early development software 3D-engine for Flash ( AlternativaPlatform , Away3D, Papervision), was supported the development of Molehill. Nowadays the most successful design is AlternativaPlatform.

AlternativaPlatform developed 3D-engine, physics engine, GUI library and high-performance server-side to create a unique browser-based games, applications for social networking, promotional and demonstration projects.
With AlternativaPlatform it is possible to create large 3D systems launched directly from your browser. For example, the currently running project development interactive designer rooms in the browser . It was also necessary to think of a scenario of user behavior, to improve the usability of the system.

Дизайнер комнат
Figure 1 - Interactive room designer in browser

Artificial intelligence


Artificial intelligence are often required for gaming systems and systems-emulators. The main categories of artificial intelligence used in such systems is:

  • thinking like a man
  • coming as a man
  • correctly select the optimal (rational) decision
  • acting rationally
Currently, there are many problems and many possible solutions related to artificial intelligence. We consider some of the problems.


Choise Problem

As indicated above, the artificial intelligence is implemented in such a way to act rationally. But rational behavior is not always characteristic of the man. People often rely on intuition. it is difficult to assess the rationality and correctness of the choice of AI, particularly at the stage of its design.


Correctness Problem

It is simple to make user the loser in the majority of games and simulations. The developers are asking: how to improve the realism of artificial intelligence , which operates correctly?
The easiest way to stupid AI is bordering its computing resources. It is obviosly that the program immediately would play worse. For example, pool have the following options: to add to the impact on the correct trajectory of the random error. But the problem is that it reduces the realism of the game. In this case, the computer starts making stupid mistakes too. Some people would never have done this kind of errors.
In other words, normal deterioration of the game AI is not very realistic . It is low pleasure to the player.
From this position, there is one solution. It needs to be done so that the computer was playing at full strength, but handed the game. As in sports like a series of small, seemingly random failures that lead to the loss or significantly increase the chances of the enemy. That is, the AI should do smart, thoughtful errors.
This kind errors requires significantly more computing resources than fair play at full strength, but the man has a real chance to defeat a powerful opponent. Developer need to demonstrate outstanding intelligence to create artificial stupidity of this kind.
To make the AI dumber should not reduce the accuracy of the calculation, but on the contrary to increase it. AI should not just be smarter than man , and the two heads of its smarter to realistically he succumb. It should to predict a person on a few moves ahead to fine-tune the perfect chance, which he should use. Or do not use it, but then lose. But you still need to give him a chance.


Lost Object Problem

Consider the following situation. There are 2 objects, one object chasing a second. Obviously, when computing the trajectories of two objects in the race, it is necessary to calculate the obstacles that come across their path, as well as scope.


Figure 2 - Object loosing

You can see that catching the object should realistically be lost in a situation where good object hiding around the corner , and this does not depend on the scope of both objects.

Summary


There are many areas to create unique behavior scenarios. These scopes have been created by new technologies. There is also a wide range of technologies, where you can use the obtained results in research, testing and modifying the model developed by the algorithm. When considering the means to study behavior scenarios in the online interactive complexes should lean toward technology AlternativaPlatform using Adobe Molehill , as these technologies are promising and have good support. It's preferred to use Microsoft Kinect and XNA Games for the desktop solutions.

Sources


1. Stage3D APIs for Adobe Flash Player and Adobe AIR - Adobe Labs
Internet-resource. URL Access: http://labs.adobe.com

2. Джоб Макар. Секреты разработки игр в Macromedia Flash MX.

3. Анатолий Ализар. Как научить ИИ делать продуманные ошибки
Internet-resource. URL Access: http://habrahabr.ru

4. Lennart Steinke. Искусственный интеллект в играх: передвижения монстров.
Internet-resource. URL Access: http://www.iskint.ru

5. AlternativaPlatform
Internet-resource. URL Access: http://alternativaplatform.com

6. David Bliss. Using Kinect within Flash Applications.
Internet-resource. URL Access: http://odopod.com

7. Елена Сагалаева. Искусственный интеллект в играх.
Internet-resource. URL Access: http://www.addconf.ru

8. Алекс Дж. Шампандар. Искусственный интеллект в компьютерных играх: как обучить виртуальные персонажи реагировать на внешние воздействия.

9. Kinect
Internet-resource. URL Access: http://ru.wikipedia.org

10. Ричард Псмит (Андрей Ленский). Что такое хороший сценарий?
Internet-resource. URL Access: http://lki.ru

11. Andrew Luppnow. Проблема проектирования искусственного интелекта
Internet-resource. URL Access: http://www.realcoding.net




* The master's work is not completed yet. Final completion is oin December 1, 2011. . Full work text and subject materials can be obtained from the author or his master after this date.