Study On Artificial Intelligence And Virtual Reality Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Virtual reality earlier was just a toy for children but now simulations are used in almost every field. For military training , Medicine, Engineering, Architecture design. Simulations are being used to prepare for emergencies(e.g. bomb blast, floods, terrorist attacks etc). And also to analyze physics of simple bike engine to complex rocket engine. For example now before a aircraft is built it should prove its prowess in simulated environments.

Virtual reality which is a young research area in computer science. Early days we saw just small simulations with limited capabilities and limited realism. In early days such limited realism was indispensable because of limitations of computational power. But now a day's computers are much more power full and advance. So scientists are thinking of inducting more realism for their simulations. But there are still some limitations in computational power. So it is possible to bring some considerable realism to the simulation under existing hardware powers and limitations

Over the years artificial intelligence has been a fascinating idea for mankind in their quest for a human like machine. Virtual reality which is striving for better realism is slowly but surely been inspired by Artificial Intelligence. AI is dominated by concepts such as Artificial Neural Networks, Genetic Algorithms, Fuzzy Logic, Ant Colony Optimization.

Many of these AI concepts requires lot of processing so if it is used in a virtual reality simulation it will eventually slow down simulation speed which will not please the users. But out of the Fuzzy Logic requires comparatively a very light weight process. Because of that use of fuzzy logic is growing rapidly in virtual reality research.

This document mainly consists of knowledge I gained by reading the research papers I found about may topic and some details about papers that describe the use of fuzzy logic in virtual reality simulations and some details about papers which describe about use of fuzzy logic in control systems.

Chapter 01.


Virtual Reality

This is a hypothetical 3D world generated by the computer. Here user can interact with the 3D world and if the realism is very powerful user will feel that he is interacting with the real world. These simulated environments are primarily based on virtual experience which is normally displayed on computer screens or other stereoscopic displays. But sometimes it supports some other sensory information such as sounds and occasionally sense of touch as well.

Interaction with virtual environments is done by many ways and means. It is usually done by keyboard, mouse, joy stick and in advance simulations devices such as wired glove and omnidirectional treadmills. Theoretically experience user gain from interacting these environments should be exactly similar to the currently the experience user gain by interacting with these virtual objects is very far from reality. This inability is occurred due to limitations of processing power of current computer systems.Virtual reality application areas

* Entertainment

* Aircraft design-(Virtual reality simulation is used by McDonnell-Douglas Corporation for F-18 aircraft design also lockheed martin use VR for F-22 advance fighter program. In addition Russian Sukhoi design bureau is using VR for their SU-37 and advanced SU-PAK-DA fighter programs .)

* Medical simulations- Mainly in surgery simulations

* Medical treatments- Mainly in phobia treatments

* Engineering and architecture

* Archeology- For reconstruction of artifacts

* Film industry

* For education purpose

Chapter 02.

2. Fuzzy Logic

2.1 History of fuzzy logic

The theory of Fuzzy Logic was first presented by professor Lofti Zadeh. In addition the term "Fuzzy" first was used by the Dr. Lotfi Zadeh.Dans in the journel of engineering " proceedings of IRE " .Mainly fuzzy logic is heavily used in control systems. This new one approach considers about not having sets with rigid borders as the conventional sets. But fuzzy logic was not used a lot practically just because of the lack of computer strength until the last years 70. Normally the people do not see the world in the precise manner. But practically they are extremely adaptive. Therefore if the systems can be programed in an imprecise manner the resultant system will be extremely adaptive and effective.

2.2 How does fuzzy logic work

Normally in a conventional system we attempt to model a system mathematically. But it is not the case with fuzzy logic. Normally FL uses a rule based if and else method to resolve problems. For example lets consider the temperature. In the conventional manner we say that if the temperature is bigger than some value we consider it as a hot reading. Suppose if the temperature is higher than fifty Celsius one it is hot or it is hot. If of this manner a temperature is 50.1 degrees it is hot but if the temperature is 49.9 it is cold. Really that is a very bad conception and this not to be flexible. In the fuzzy approach of logic use us terms as if (this is too fresh) AND (the process obtains colder) THEN (heats the process) and it brings us more better result

Normally in a conventional system we attempt to model a system mathematically. In contrast fuzzy logic works different than conventional problem solving methods. Normally FL uses a rule based if and else method to solve problems. For example lets assume temperature. In conventional way we say if the temperature is greater than some value we say that the area is hot. Lets assume if the temperature is higher than fifty Celsius it is hot or it is hot. So in this way a temperature is 50.1 degrees it is hot but if the temperature is 49.9 it is cold. Really that is a very bad design and it is not flexible. In fuzzy logic approach we use terms like if (it is too cool) AND (process is getting colder) THEN (heat the process) and it brings us better results

2.3 Fuzzy Linguistic variables

In fuzzy logic these variables are called nouns other than these variables are treated as words rather than numbers. Normally input category is a noun eg:- "temperature"," price" , "altitude" etc. Error is the difference between given output and expected output. Error is also stated in the same way. In fuzzy logic we use terms like "large positive error" , "zero error".

2.4 Fuzzy rule matrix

Fuzzy linguistic variables can be effectively used in control systems. Earlier we saw that fuzzy parameters of the error can be modified with adjectives such as "negative", "zero" and positive etc . First we need to map the universe of all possible inputs to the system. In this example simplest solution is to use a 3*3 matrix. Columns of this matrix is names as "negative error" "zero error" and "positive error". Rows represent the negative positive and error rate inputs. This matrix is called rule matrix. Here we have nine possible logical AND products output of the matrix.

1.5 Fuzzy membership functions

After we created the rule matrix we need to apply these rules to the system. This is where the concept of fuzzy member ship function comes in to play. In membership function is graphical representation of magnitude of each input. It is used to determine the output. In addition it shows functional overlap between inputs ad well. Graphical representation of a membership function has some shape. Some are triangular. It is the most common shape. In addition bell. harversine and exponential shapes also can be seen.

Infernecing with fuzzy rule base

In the rule base we can see lot of rules . We need to infer the logical product of each of these values. So several inference methods are used.

1. Max-min rule

This method tests the magnitude of each rule and selects the highest value

2. Max-dot rule

In every membership function take the horizontal peak value in threspectiveposition and get the composite of horizontal peak area under each function.

Chapter 03.

3. Use of Fuzzy logic in VR environments

Artificial intelligence is a very powerful techniques which gives any application powerful human like capabilities. Normally intelligent behavior make any application very smart and adaptable to any environment. Now we need adaptive smart application which can response well unpredictable situation. in such scenarios artificial intelligence is very vital.

In a virtual reality simulation artificial intelligence can play a major role. Because the simulated environment in the virtual reality simulation need to be very close to the real natural environment. For example if an agent in virtual reality environment need to navigate in the virtual environment it should be smart enough to reach its goal by evading all obstacles he faces on the way. In these situations artificial intelligence can play a huge role

There are many approaches of artificial intelligence. Such as Artificial Neural Networks, Genetic Algorithms, Ant Algorithms, Fuzzy Logic, Hidden Markov models and many more. Biggest problem with these artificial intelligence approaches is that they require lot of processing power. For example if a Neural Network is used for some process in real time virtual reality simulation it will need lot of time to the Neural network to train and produce the output. So Artificial Intelligence approaches which will need very heavy weight process is very difficult to be used in real time virtual reality simulations. So if we are using Neural Networks we will have to train the Neural Network and use the trained ANN in the simulation.

But in comparison to the other Artificial Intelligence approaches implementation of fuzzy logic can be achieved by relatively very light weight process which is a significant plus point over other Artificial Intelligence approaches. So because of being very light weight Fuzzy Logic can be used in real time Virtual Reality simulations. For example if two agent in a Virtual Reality environment fight with each other we can use fuzzy logic to make their more realistic.

Because of being very light weight use of fuzzy logic in Virtual Reality research is increasing day by day.

Chapter 04.

Fuzzy Model for virtual agent controll in virtual reality environments

In fact every entity in virtual reality simulation can be considered as an agent in virtual reality environments. But because of computer performance reasons every entity in a virtual reality environment is not modeled as entities. This research done by S. Vosinakis of Department of Informatics University of Piraeus proposes to Improve the performance of these virtual agents using fuzzy logic controllers


Fuzzy logic is already used in electronic control such as electronic motor control systems. So fuzzy logic has the potential to be used in control system for virtual reality agents.

In Virtual Reality simulations a virtual agent is considered as an autonomous entity in the 3D space. These agent s are considered as synthetic characters which interact with their virtual environments through their virtual sensors. For example a car or some kind of animal in our simulation can be considered as an agent in our Virtual Reality simulation. Just like their counterparts we see in our day to day life these virtual agents also shows some complex behaviors. So researches has made these agent intelligent so they behave smarter

There are many approaches for these agent control. The simplest approach is to use scripting. Here for control purpose lot of "if and else " statements are used. This is the most simplest and widely used method. Main advantage of this scripting approach is its good directional control. IMPROV is such popular scripting system. But the biggest disadvantages of this approach is its inflexibility which means we need to predefine everything.

In addition we have sensor driven control which is derived from AI research. Here incoming stimulus is coupled to ongoing reaction. These agents are usually equipped with an AI planner. This AI planner uses symbolic reasoning approach and takes decisions according to agents intentions. In relative to scripting approach this approach is much more flexible in changing environments. But the problem in this approach is its reasoning on which based on symbolic description of the environment. This description is stored in global database and changes according to predefined effects of agent's actions. Although this approach is relatively flexible than earlier discussed scripting approach this approach also have the problem of lack of flexibility. Here the problem is in some dynamic environments the effects of some actions may not be known a priori(e.g. sports such as football), In these situations we need a spatial reasoning engine that can reason about object relations in a higher level and dynamically update the world's symbolic representation.

Fuzzy logic is effectively being used in electronic control systems. Later it is used in robot navigation and obstacle avoidance. This paper presents way of using fuzzy logic to define inter object spatial reasoning. This paper introduces a variation of fuzzy sets for the especial case of the 2D plane called Fuzzy Region. It also defines a respective rule-based system for dealing with spatial problems in Virtual Reality environments

4.1 Fuzzy Regions

This concept is based on the concept that using fuzzy logic we can represent spatial relationships like "near", "in front of", "beside" which can be represented in fuzzy sets with some degree of member of. Here fuzzy region is considered as a special case of fuzzy set. Here universe disclosure U is a 2D area and membership function gives us the membership degree of a the particular point in our region. Here the universal disclosure(U) can be continuous or discreet.

Figure 1.1

Here the area is shaded according to the membership values. By using these fuzzy areas we can define fuzzy set operations.

4.2 Proposed fuzzy rule base

The fuzzy regions are used to construct the fuzzy rule base. Fuzzy rules have two parts.

A premise consists of one or more antecedents

A conclusion consists of one or more consequences

Here antecedents have following form. Reason for using a technique like this is that the agent logic cannot be based on whether some point is in an area of interest. It really depends on whether the whole object satisfies such condition

4.3 Object region IN Fuzzy Region

This value will be calculated by getting the mean of membership values of the points of the whole object

Then these antecedents are then combined in to premises using NOT, AND or OR functions. Results are also Fuzzy sets which are assigned with some output variables. A consequence of a rule may have several consequences. A fuzzy rule fires in to some kind of degree depending on belief levels of each antecedents of the premises. These antecedents are evaluated using membership functions which produce belief levels. These belief values modifies the fuzzy output region.

The conclusions of fuzzy rules are then combined in to the final decision. Then each fuzzy region should be defuzzified to produce the final result here a 2D point. Here centroid and the average of maxima. In the first case we calculates the centroid of the volume defined in the fuzzy region and projects it on the 2D plane to find the defuzzyfication point. In the second case we find the average of the points in the region that have the maximum membership value. There is another third method which is suitable mainly for navigation. This is called nearest maximum method. This method returns defuzzyfication point. This point is the maximum membership value in the region and also it has nearest distance to a given point or reference. So this method can be applied to agent's current location as a point reference. This will return the nearest target point which satisfies the fuzzy rules. Depending on the problem the defuzzification method could also be used in a smaller region around the reference point and not to the whole universe of disclosure.

4.4 Implementation

Instead of using arrays for fuzzy 2D space use of ordered set of points are proposed. The system assumes that there is a linear connection between each point to next. In addition al points except first and last , takes the value of their nearest extreme. Here a fuzzy region is defined by X and Y fuzzy sets. Whenever a membership value of some point is to be calculated �X and �Y value of particular set is calculated. Then they are combines to produce the finale to the finale value. Although this method is less computationally expensive

but this method restrict the region to some form. We can't declare arbitrary points.

Chapter 05.

A Fuzzy Action Selection Method for Virtual Agent Navigation in Unknown Virtual Environments

5.1 Introduction

Behavior based control is dominant part in agent control. This paper presents a new Action selection method based on fuzzy alpha levels and Huwicz criteria.

5.2 Background

In a behavior base d system behaviors are considered as processes to achieve main goal of the agent. Here the goal of the system is achieved by subdividing the overall tasks by small subtasks. Normally a action selection method computes which action should be executed by BBS.

Normally action selection methods(ASM) classified in to two groups. That is arbitration and fusion. Arbitration ASM allow one behavior or set of behaviors at the same time to take the control for a period of time until another set of behaviors is activated. This process can be one of the following

* Priority- Action is selected by a central module based on a priori assigned

* State based - Select a set of behaviors which are competent to handle the situation

* winner-takes-all - Set of behaviors compete with each other behavior with maximum response will take over the control.

* Voting- The action with maximum weighted sum is calculated by considering the output of each behavior

* Fuzzy - Similar to voting but uses fuzzy internecine mechanism

Here Fuzzy Logic is used to this purpose. Architecture of fuzzy controller is comprised of three behaviors that is path planning, goal seeking and obstacle avoidance

Overall architecture

IF x is Y AND y is B THEN is z is C

This is a fuzzy rule where x,y,z are linguistic variables. A fuzzy associative memory (FAM)is used as a process of encoding and mapping input fuzzy sets into fuzzy output sets. For example a set of fuzzy rules R=R1:R2:::::Ri:::Rk.. The rule Ri is defined as follows

If X1 is A1m AND X2 is a2m and::: amd xN is Amn THEN Z is Cmn.......(2)

The following relation will implement Ri


We can rewrite equation (3) below

Ri(X1:X2::::::Xn:Z)=[A1m(X1) ^A2m(X2)........ ^]>Cnm...............(4)

Where X1:X2.............Xn are input variables which are sensor data of the virtual agent A1m:A2m::::::::Anm are the input fuzzy sets. Cnm is the output fuzzy set Z is the output variable, n is the dimension of the input vector and m is the number of fuzzy sets

In order to create an n fuzzy input vector X =X10:X20:::::::::::::::::::Xnm.......(5)

The system needs to compose the input vector X with the calculated fuzzy relation Ri to produce the following input C

Where Xmn is the fuzzy crisp value Xmn into fuzzy output class Cj(Z). The output of the ith rule

The system needs to compose the input vector X with the calculated fuzzy relation Ri to produce the following input C

Where Xmn is the fuzzy crisp value Xmn into fuzzy output class Cj(Z). The output of the ith rule

Ci=[ A1m(X1) ^A2m(X2)........ ^]>Ci

Here mamdani method is used to defuzzyfication

C= SwiCii

C= SwiCii[A1m(X1) ^A2m(X2)........ ^]>Cnm

Wi is a non negative value.

Defuzzyfication response= SwiCii/Swi..........................................(6)

Equation four and six used to derive Fuzzy Action Selection method

The action selection method

This method uses fuzzy alpha levels and fuzzy subtraction operation to calculate the area of a new fuzzy number. Here it is produced by the comparison two fuzzy numbers. This research uses a reduce redundancy of this calculation.

Let �X~(X) be the membership function of a fuzzy number X~ behavior output defined on R. Here n assumption about the normality of �X~(X) are made based on the left and right of the fuzzy alpha cut of the fuzzy number, X~, are �L X~ a(x) and �R X~ a(x) and 0< a<h X~.

Here h is the height.

c and d are at the minimum value of the left and right spread of all fuzzy numbers. Interval subtraction is used to simplify the subtraction between fuzzy numbers


Let X~1, X~2, X~3............... X~m be m arbitrary bound fuzzy numbers produced by each behavior.

1. Set the height hg(x), common maximizing barrier and c for referential rectangle R~

2. Determine the subtracted interval numbers

3. Determine behavior weight for each fuzzy number

4. Repeat step 2 and 3 for every interval

5. For every W use the minmax(maxmin) criterion which select the lowest value for fuzzy numbers

6. Determine the index optimism a. Then the finale behavior is selected by the huwicz criterion

Chapter 06.

Modeling human behavior at work using fuzzy logic

6.1 Introduction

Correct selection of individual workers to tasks is a very hard task. So AI can be used to simulate the human behavior. This research paper propose a Fuzzy Logic based method to represent the people and some of the people characteristics. This research focus on the fuzzy characteristics and how the team behavior can be modeled with agents(people) and project characteristics.

6.2 Modeling human capabilities at work

Modeling human behavior has been a major problem. Because human are unstable, unpredictable and capable of taking their own decision. In addition in a working environment performance of each individual will vary depending on their experience, ability, education and their current physiological states.

This paper suggests that there are three challenges in human behavior modeling

* Human are not limited to one identity or any common set of emotions

* Human are not limited to acting in accordance with predetermined rules

* Human are not limited to acting on local patterns

As simulating human behavior as a whole is almost impossible this paper proposes that it is possible to simulate some parts of human behavior. Here this paper proposes to simulate behavior of a working team. First thing is to identify the set of relevant human characteristics that affects the performance of person in a team. These characteristics can be groped in to four categories

* Cognitive capabilities- This involves complex brain processes

In these paper these cognitive capabilities were defined by the degree of expertise of some individual in some field.

* Personality trends- CLEAVER technique is used for this purpose(This is a questionnaire which give a numerical value about personality trend parameters)

* Emotional trends- This research look at two trends that is positive emotion and negative emotion

* Social characteristics- Here characteristics such as good communication and co-ordination is considered.

6.3 Fuzzy logic to model human behavior

After parameters of internal characteristics are measured, these characteristics are represented and measured as follows. All the characteristics are linked together. The behavior of a person is generated by combining above mentioned characteristics.

But it is intriguing that a simple numerical value can represent human emotions and

Complex behaviors. Here three qualitative attributes are used they are low, medium and high.

Fuzzy logic can be used to model this three characteristics.

6.4 Identification of fuzzy parameters

First step of this model is to identify fuzzy parameters. And also their ranges need to be found. Fuzzy sets are used to parameterize the main aspects in the modeling which includes following

* Agent internal characteristics.

* Tasks

* Agent performance

* Modeling of the human behavior

6.5 Internal characteristics

Internal characteristics are fuzzified by using a Gaussian membership membership function. For emotion, cognitive and social characteristics three intensity fuzzy sets are defined. Range of the values of these fuzzy sets range from 0 to 100.

Low intensity- 0 to 35

Medium intensity -25 to 75

High intensity -60 to 100

In addition fuzzy sets were also used for

Increase/ decrease fuzzy sets also defined for the emotion intensity. It is calculated as a result of as a result of firing behavior rules in the simulation process.

Fuzzy sets used to increase/decrease the emotional and stress intensity

6.6 Task parameters

The behavior of the work is modeled through interaction between team members. Tasks modeling setting values to 11 parameters.

1. Number of participants in the task.

2. Estimated duration (measured in days).

3. Sequence (sequential or in parallel).

4. Priority within the project.

5. Deadline.

6. Cost.

7. Quality.

8. Application domain.

9. Task description

10. Difficulty.

11. Type (required specialization level).

Last two parameters are fuzzy parameters. This parameters shoe required specialization level to achieve some task.

6.7 Agent performance parameters

Following parameters are proposed to used to measure the performance of the each individual agent in the model

1. Goals achievement.

2. Timeliness.

3. Quality of the task performed

4. Team collaboration level.

5. Contribution of each individual

6. Required supervision level.

The researchers has used this three parameter because these are the parameters used by project leaders of some petroleum companies. Here also the performance parameters value range from 0 to 100. In is divided in to five fuzzy sets. Very low(0-30),minimum(45-75),acceptable(65-95) and satisfactory(90-100)

Fuzzy sets to model human behavior

In this research fuzzy rules are used to simulate how workers might perform in some situation. The fuzzy values which fire from -20 to 20 modify the timeliness parameters. The fuzzy values of them are high_advance(-20 to -5), medium advance(-10 to 0), normal(-5mto 5) meduum delay(0 to 10) and high_delay(5 to 20). When all these values are deffuzyfied crisp values are calculated and assign to corresponding parameters.

Fuzzy rules for modeling human behavior

After fuzzy parameters are identified and defined, it is needed to build the fuzzy rule base required to build the simulation. In this research it is done three steps

Here there are three sets of fuzzy rules are used.

1. Fuzzy rules to modify the agent internal state

2. Fuzzy rules to get the agent performance

3. Fuzzy rules update the agent internal state.

Modifying the internal state of the agent

In this research the project manager selects initial set of possible team members. Also assign tasks to each members. After that the simulation begins. Then emotion and stress values are set for the internal state of the agent. For the research it is assumed that

All team members have medium intensity values for the emotions at the beginning. According to the internal and external factors the corresponding fuzzy rules are triggered. intensities of the agent's emotion and stress are modified in the simulation.

IF T1 presents a high_delay AND A1 has a driver personality with high_intensity


The desire emotion will have a high_increase

The interest emotion will have a high_increase

The disgust emotion wills stay_equal

The anxiety emotion will have a low_increase

The stress will have a low_increase

IF A1 is introverted AND in T1 must interact with other people THEN

The desire emotion will have high_decrease

The interest emotion will have a low_decrease

The disgust emotion will have a high_increase

The anxiety emotion will have a low_increase

The stress will have a low_increase

Generating agents performance

Nest set of fuzzy rules involves the modeling of agent performance. Because of that the rules settings the agent performance parameters for each assigned task are triggered.

Given the agent A1 in charge of task T1,

IF A1 has a high creativity level; A1 has a driver personality with high_intensity AND

T1 requires a high specialisation level THEN

The goals achievement is normal

The timeliness has a medium_advance

The quality has a medium_increase

The team collaboration level is normal

The individual contribution has a medium_increase

The required supervision level is normal

IF A1 has a low experience level AND T1 is a high difficult task THEN

The goals achievement has a medium_decrease

The timeliness has a high_delay

The quality has a medium_decrease

The team collaboration level has a medium_decrease

The individual contribution is normal

The required supervision level has a medium_increase

6.8 Implementation details

JADE framework is used to build the multi agent system

* The software agents do not work to solve any real project but they only simulate

their interaction with other agents and with their assigned task(s).

* A plausible set of global behaviors of a team is obtained by averaging its behavior

over a statistically significant number of simulations.

* c) The most suitable team configuration can be obtained by comparing the sets of

global behaviors for several possible team configurations.

* d) We cannot foresee the future, so we cannot guarantee that the team will behave

exactly as the simulations suggest, but we aim to generate information about

possible performance patterns. This information can be particularly useful in the

identification of undesirable performance patterns and their relation to the team

configuration and task assignment.

Chapter 07.

Fuzzy Logic based method for virtual surgery simulations

7.1 Introduction

This research is based on using fuzzy logic for surgical simulations. As this is a real time process Fuzzy Logic can be effectively used.

7.2 Design of Fuzzy Logic system

Cutting of a virtual surgery depend on two parameters force on the virtual tool

and stiffness on the virtual tissue. These parameters are fuzzified to get membership values corresponding measured parameter values. Then they are fed to obtain fuzzy rule base

Defining fuzzy membership functions

Fuzzy controllers based on the experience of the expert inspired by real people. The functions for membership Inputs and outputs must be defined by the experience of experts. The membership function defines the fuzzy sets in each input and output variable. In our virtual surgery system, the membership functions defined for the input variable (force and stiffness) and the output size (depth) as shown in Figure 2. Each member Ship function is divided into several fuzzy regions (small, medium, etc.). The x-axis of Figure 2 (a) provides Input force and the normalized y-axis quantifies the partial membership values of a particular group in any Fuzzy region. The x-axis of Figure 2 (b) represents the normalized stiffness of the soft tissues while the y-axis Teachings of partial membership values of a given stiffness in each fuzzy region

Constructing fuzzy rule base

The fuzzy rule base transforms the input functions given products. To create the rule base, a

Correspondence table is defined. The lookup table defines the appropriate actions are taken for each combination Input fuzzy sets. To achieve this goal, a three-dimensional array is constructed. The input functions are represented in the axes X and Y. The output function is represented in the Z-axis in the virtual surgery system

Input function of the force in the x-axis and Y-axis stiffness of the output function of depth

represented in the Z-axis, the reference table for virtual operations of this fuzzy system is given in Table 1.

Membership function for force factor

Membership function for stiffness

7.3 Appling Fuzzy rules

The membership value of each fuzzy set in the input functions determined. It is calculated from

Definition of the function member. For each combination of input membership values, the combination of fuzzy logic determined by fuzzy mathematics. This is because the rules we have written rules. All values each category of members are collected and added output. De fuzzification is achieved by the moments about the origin, including all members Output Values. This method is commonly referred to as the centroid method of de-fuzzificatiion.


The fuzzy virtual surgery system has been used to make the cut tissue (Fig. 3) on the basis of existing

the user and the material properties. The experiments were conducted to determine the validity of the estimate of the fuzzy


Chapter 08.

Fuzzy logic base 3D collision detection algoritm 3D collision detection

8.1 Components of a surgical simulation

Simulation cycle of proposed surgical simulator

8.2 Proposed collision detection algorithm

We define two vectors representing the state of the scene. First, the motion vector normalized VM the direction of movement of the tool on the Body between two steps a simulation. Secondly, a vector represents the surface of this body. Collisions may affect several vertices the deformable model. It was adopted in the preliminary working with all the corners to deal in conflict, because the facet

Crossings can be prevented in this way. However, if the Collision of two facets oppsosite the tool, all corners move forward together to achieve one of them and the tool is. This effect may appear in scenes of the simulation as the basic cutting. So independent vertex manipulation has been accepted. Each outer normal the facet j around the corner i must be considered,

. These two vectors are normalized. In addition, three cases of interactions are distinguished,

all aspects of around a fortress Vertex, the system is fuzzy feedbacked. The outputs of the system

Tuple I = (IP, IS, IE), weighing the degree of similarity the situation on a case by case interaction, and it is also an application input to the next iteration. Therefore, the displacement vectors

every corner collided deformable model is obtained as follows. Finally, Figure 2 shows the membership functions IC linguistic variables and the tuple I. It is evident in this figure. Here its S represents a threshold between the interaction Cases. These cases are considered equally likely

(a) Fuzzy membership functions

(b) Fuzzy linguistic variables


Robotic soccer with Fuzzy Logic

9.1 Introduction.

Traditionally, we think of robots as a very independent individual They interact with their environment. On the other side of the robot's environment may be another robot, and if there is any kind of communication between robot interaction must be social. In this case, we see the robot system as a group of individuals, and take care of their individual emergence behaviors. Good tool for development and testing of individual behavior and social robots is robot soccer . It allows easy generation of different situations and comparison of the effects of their control mechanism by playing against each other. Robot soccer is a game between two teams of robotic soccer players. Subject to the

the controller is a robot soccer team with five members. A two-wheeled robot that transport equipment, radio transmitters for communication and 16-bit controller used for basic control of electric motors Micro (Fig. 2). The robot does not even does not recognize the input element, is the only entry in the robotic system , camera with image recognition (Fig. 1). Part of the system is also home PC

machine for high levels of control and image recognition

Fig- 1

Fig - 2

9.2 Fuzzy expert system

Expert knowledge is often dealing with uncertain knowledge. This special

applied to information in the field of robot soccer, where we wok with terms such as

"Close", "left" and the like. Fuzzy logic enables us, with this type of work

Information. Soccer game contains more than one person by definition, In robotic soccer there are more robotic players and each has its own task and requires different approach and knowledge base.

It is clear that we do not think the football team, just how to move object. Multi-agent theory allows us to this problem in several organizations and dissolve. Think of all the parties separately relevant knowledge. Design of multi-agent system consists of several types of agents together. Each it was his role in the decision-making. Their independence enables a good decomposition of the problem and secondly, working together, again creation, and their contributions to control up to a complete system. In the multi-agent system control robot football we need more types of reasoning and agents a kind of broker-agents have to ensure communication, synchronization Information and caching. Definition states that the autonomous agents in a system and is Part of an environment that believes that the environment and acts on it over time pursues its own agenda and give effect to the direction in the future. Our Cases, the autonomous agents the agent with fuzzy expert system for reasoning

General information about broker agents can be found in. However, in this

work we use term 'broker agent' more intuitively. Purpose of broker agent in our apprehension is to collect data from other sources and provide them for other agents. They do not perform any reasoning by themselves.

9.3 Components of the system

We have stated following definition: System using multi-agent distributed architecture, expert system and fuzzy inference mechanism we call multi-agent. When we are looking for models for multi-agent system for soccer robots have control, which was ideal for the true football game created by real people football match together. The game is packed with players on the field, but even coaches, referees, spectators, even those who work in television transmission it allows us to see the game at home are involved throughout the process. Taking

given these facts, our tax system contains the following entities: agent players Agent coaches, agents and brokers television postman.

Player Agent represents a person that has contact with the ball. It emulates human abilities to get to the best position, acquire ball, pass to other player or shoot to the

goal. Player can play role of goalie, attack or defense. Player's ability to see in Player Agent is implemented as Visual Module. Purpose of the Visual Module is to receive absolute coordinates from agent Cameraman, transform them into radial form and insert fuzzified information into inference system. Every object, visible to the Player is transformed into radial coordinates which are more close to human perception.

Ability to receive messages to other players and officials can entrainer compared to the abilities to hear from real people. This kind of agents can be used for input take the intentions of the mind and the states of other agents. For instance, if robots the ball can be and it announced the rest of the team can remember his announcement and focus on obtaining better positions in the court received Agent of the trainer. Strategic storage for the states of the game. We can use it to recall the data not that with each turn of the decision cycle, such as intentions of the robot to change its The strategic objective relevant ads or other agents. In contrast to experts Database system to maintain the memory contents of the strategic game

Assessments reflect the strategic situation. Memory contents are changed from strategic receiving messages from other player (someone from the team received the ball), or if the change of Game situation is realized in the justification process.

Player's skill to play soccer is apprehended as ability to process visual information and resolve it into action. When we take visual and hearing input, reasoning using player knowledge should lead into deducing desired player's action. This action is then executed by the inference system and speed vector is produced. This vector then can be sent to other agents for further processing

Each player also has a memory that contains critical information about current games

Situation and policies of the last lap. Memory content is added strategic inference engine. This step is called perception. Then takes the final step system resolution information from the database and included Action decided that the player has to fulfill. The decision may be deterministic, where Actions must occur at a time, or stochastic, where the player can choose

several possible actions. Stochastic decision may be influenced by the control

parameters, then we can set the preference for certain measures. These parameters can

individual characteristics drive express (team game, agile ...). Part of the Player's

knowledge base in terms of resolution step is like a tree of the decision represented on

Image. Last step is called execution. In this step, it will be the resulting action and its target (ball

or the center position) is taken in order by means of fuzzy inference Madani, speed also obtained. The vector is the result of the inference process together.

Agent coach- Stand as the director of the team. These agents does not play the game but focus on the strategy of the team. It sees the playground in birds view. Here agent's visual module takes absolute co ordinates. Coach use fuzzy linguistic variables like "penalty area" "center of the field" etc. These linguistic terms insert in to he inference engine. Which result in recommended positions for the agent type "player".

Agent cameraman- This agent sends a message with the coordinates of all objects visible

All employees, which contain a kind of perception of visual information. These messages can also serve as synchronization signal for the whole System. Whenever the autonomous agent is replaced by new visual information, it begins a new cycle can run in its reasoning.. Purpose of agent cameraman is the visual information to the playground for agents who need them. It has no decision has skills, his goal simply caching and delivery of visual information.

Agent postmen - Postman is only acts as a buffer to collects information on the results of the decision taken by the players. It processes and sends out synchronized with the outside world. But this agent is very simple and contains almost no logic


Implementation of this was done using JADE agent framework

10. Conclusion

As discussed in this document Fuzzy logic is making huge impact on virtual reality simulations. First paper talked about using a special fuzzy set called fuzzy 2D region. Second paper I discussed which also talks about virtual agent navigation use fuzzy alpha values and Huwicz function. Considering the overall performance in the two approaches the method use fuzzy alpha values is much efficient because it significantly reduce number of comparison which results greater efficiency.

Then I discussed about using fuzzy for human behavior modeling. But according to the research they have put lot of limitations and assumptions in their research. In addition many complex human behaviors and complex human brain activities were not yet fully understood(It is interesting whether a human brain can understand itself!). So I don't think that modern computer science is matured enough to model human behaviors. But Fuzzy Logic has certainly has shown lot of promise in this area but still there are lot of limitation and assumptions which make existing research models place themselves very far from usable human behavior model. I think lot of research and advancement of computer hard ware will give more promise. In addition only fuzzy logic may not be enough for this purpose. Fuzzy Logic in collaboration with Artificial Neural Networks will be more effective. May be new AI approaches such as new Ant Colony Optimization(ACO) and Artificial Immune Systems also can be considered

Next I discussed about use of fuzzy logic in surgical simulations fuzzy logic in this area has shown lot of promise. There fuzzy logic has shown lot of promise and maturity. Next I discussed how fuzzy logic can be used in robotic soccer. Fuzzy logic which needs very light weight process has shown lot of promise in this area.

As a whole I think for real time simulations fuzzy logic is the ideal Artificial Intelligence

technique available today. Because it not only heavy enough to produce good AI it is also light enough to fit with modern hardware. So Fuzzy Logic will be heavily used in neat generation VR research and simulations.