ࡱ> JLGHI5@ bjbj22 PXXVoppp8*q>tDhuu(uuu$LREj(^uu4\0&uuR:M,շuu p`M4pGy TR>ͯ>շ>շ vbؖ#dJ8?$1J? COGNITIVE ROBOTICS AND ROBOT PATH PLANNING Mladen Crnekovi Mladen Su evi Danko Brezak Josip Kasa Prof. Mladen Crnekovi, PhD, University of Zagreb, FSB, I. Lu ia 5, 10000 Zagreb Mladen Su evi,MEng, University of Zagreb, FSB, I. Lu ia 5, 10000 Zagreb Danko Brezak, MSc, University of Zagreb, FSB, I. Lu ia 5, 10000 Zagreb Josip Kasa, MSc, University of Zagreb, FSB, I. Lu ia 5, 10000 Zagreb Keywords: cognitive robotics, space perception, mental space ABSTRACT This paper investigates how a human percieves its space, both in a static and a dynamic sense, and how he makes a decision of movement. Instead of precise numerical values which describe the space metrics, the mental space is established and described with identification properties of obstacles and their relationships. Special agents are activated for finding specific properties. Complex properties are being built from simpler ones; therefore the space model becomes more and more abstract. It is supposed that if we have this kind of model and goal of the movement, it is possible to make a movement decision. Only one movement step is realized in this way, while several forward steps could be planed. This permits flexible adaptation to the current space situation. 1. INTRODUCTION In the several last years robots philosophy has significantly changed. The development of all scientific areas has a great influence on robotics and its perception in public. Two decades ago the main topics in robotics were about the executing robot level, i.e. the level of motors, sensors and microcontrollers. Each robot system was studied separately, and there were few connecting points. Methods of classic automation were used for the robot control and the robot programming was very complex. The robot interface to a human was not adequate and in most cases it consisted of text exchanging by the robot control unit. In that kind of environment it was almost impossible to program the human like behavior. The aspiration to build a robot similar to human, not only by its look but also by its behavior, has introduced into robotics new disciplines, unthinkable before. One of very important discipline is psychology which answers the questions about the human behavior. In addition, it has been noticed that the writing of algorithms in the conventional sequential way is not suitable for the problems of complex nature. Fortunately, additional tools for decision making have been developed: fuzzy logic, artificial intelligence, artificial neural networks, visual recognition, voice analysis, parallel programming, system of agents, etc. All these tools have brought the cognitionrealization that we have elements to build a system (robot) with human behavior, recognized by its cognitive abilities. That kind of robot would pass the Turing test without any problems, and with some cosmetic touches it would be difficult to distinguish it from a human. Of course, this fact faces us with moral, ethic and other dilemmas recognized by science fiction writers a long time ago. These questions are not less important than technical problems and it would be very interesting to discuss them. In these circumstances maybe the word "robot" is not adequate because it reminds us of something inhuman (in mental sense). If something looks like a human and behaves as a human then it must have an adequate designation (remember a boy in the movie "Artificial Intelligence", he was looking for love). One solutionpossibilitiy is to divide the term man into two terms: bio-man and mecha-man. No matter what its name and purpose are, this creature would certainly have the movement ability, and its movements would be human like. It means that a mecha-man would have the ability to plan its movements and to reach its goal without collision with any obstacle. These movements would be optimal, i.e. the shortest, least energy demanding, least hazardous etc). In order to plan and realize that kind of motion, it is necessary to understand the human perception of space and movement. 2. PREVIOUS WORKS Previous works include a wide range of problems connected with cognitive behavior. Some of them investigate a specific part of the problem, and some of them try to understand a wider context of cognitive robotics. Some works are oriented to analysis, some to synthesis. A comprehensive review of the cognitive robotics is presented in work [1]. A humanoid robot has a task to react to its environment like a human. A process that makes it possible is called Human Agent. Above Human Agent is Self Agent, which gives the robot the cognition of itself and instigates its interaction with its environment. The reaction of the robot starts with the event when somebody approaches it. The robot will interpret it as a wish for communication. If another person approaches the robot, while it has its attention focused on the first one, the robot notices that and makes a decision about which person it will pay attention to. If the new coming person has a higher priority, the robot will make excuses to the first person, and after having completed the interaction with the newcomer, it will refocus its attention to the first person. In order to realize this ability and other robot abilities it was necessary to build models for some aspects of human behavior. The introduction of "attention index" has been suggested in work [2]. According to a realized and expected movement, the adequate event for focusing attention on the process will happen. In that case a path is successfully generated in the presence of dozens of obstacles. Work [3] pays additional attention to "emotions", and models them by a specific neural network. Work [4] uses the Q-learning algorithm to evaluate a robot learning system. The dynamic design of the space model by means of the vision system and autonomous agent is the subject of work [5]. How it is possible that even creatures with primitive mental/computing abilities (like insects) solve the problem of goal reaching very efficiently, and what is a minimal set of information to solve this problem, is the subject of works [6] and [7]. A detailed model of autonomous vehicle driving is presented in work [8]. For that purpose the problem has three major processing components: (1) sensors information processing measurement, recognition, classification, agent definition; (2) internal world reproduction processing situations, events, evaluations in a task context manner; (3) behavior generation from the world model selection of the next step to accomplish the mission goals. The whole process is based on the idea of independent agents which are switched on/off as required. Work [9] is about the integration of different world models and the ways how to describe them. This work also deals with the basic idea of independent agents called specialists. Although every paper describes a specific part of the problem (and solution), there is a consensus on the correct approach to reaching the goal. 3. HOW DOES A HUMAN PERCEIVE THE SPACE AND ITS CONTENTS? All of us are capable of incredible complicated actions, dependant on and limited by a large number of factors. When we plan a path to some destination, we travel in our mind along roads, passing squares, parks, moving beside buildings, crossing rivers and finally we come to our destination. First, we pass the whole path in our mind, and then we start a real trip. When we come across some unpredictable obstacles (material or non material) in the course of the trip, we can find an alternative solution very quickly and arrive to our destination safely. In our minds we have plans and space relationships of small areas (room, house) but also of middle and large areas (quarters, cities etc.). We can change the level of details which we remember and this is the point when we realize some limits in our memory. Could you remember how many thin laths there are in your neighbor's fence? Are the public lights in your streets round or square? Are there three or four trees in front of the bank? You don't know? Never mind. For the type of decision you must make, this information is not relevant. One look is enough and all these pieces of information will be available for you, when this information becomes necessary (for some other purpose). The previously described mental journey was carried out according to our world model called "mental space". This mental space is our individual model of the world, a picture in our mind which covers some aspects of the "real" world. The mental space is our subjective perception of the world and we change (build) it with time. The mental space is available even if we do not have visual or physical contact with the environment (for example if we close our eyes). Details level of the mental space is changeable and we can focus it. Also, there are sub models of mental space which are more detailed but also short-lived. Figure 1 shows a global plan of the process for movement decision making.       Figure 1. Phases of movement decision making Each decision level is specialized for specific tasks. There are numerous interactions between levels and all of them work parallel. Data processing on any level and function is a job for agents, i.e. independent processes (programs) which find a specific part of a solution. The number of active agents and their priority depend on the task complexity. Every time we need some specific information, a new agent is activated to perform the desired information. After the task is successfully completed, the agent is deactivated. The main problem is to decide which agent has to be (de)activated at a particular moment. The picture analysis and object recognition start with taking a picture from a camera. This is an interaction point of the cognitive system with an environment. This level of data processing is very important because without good world recognition it is not possible to build a quality space model. Mental space building is an iterative task. The building of mental space is not defined by the "real" state of the world; rather, it is defined by our recognition of that world. That is why the recognition process from a picture is so important. Since path planning is based on our mental space it follows that the model of that reality is between our decisions and reality. In other words, our decisions are results of the "picture of reality" (i.e. mental space), not of the reality itself. If we could look at the world in infrared light, our interpretation of the world could be different. This would have an effect on the planned and realized motion. In the path planning one or more parameters are optimized. But one demand is always present in every case: during the motion, collision with any obstacle in the space must not happen. With a movement in the environment we close a loop of interaction with the real space. But in fact, we can not see the movement in the real space what we see is only a picture of this movement in our mental space. As all decision levels work parallel, there is no need for some synchronization between them. But it is important that each cycle time is short enough to produce an illusion of continuum. Basic properties for the description of objects/obstacles in our mental space are: distance direction size visibility movement specific details access collision danger possibility of going around Human distance perception is linear to a certain degree. A double distance is perceived like that only to a certain limit. After that limit we do not have good proportion of the distance. This limit of linearity is individual and depends on personal experience. We can all imagine a distance of one meter because we can simply experience it by showing it by arms. One hundred meters is simple to imagine because we can see that distance. Ten kilometers demand good imagination (we would say it is from the place X to the place Y). One thousand kilometers could bee imagined only by professional drivers and pilots. No one can imagine one million kilometers, it is just a number. If we take into consideration only a part of space that is within sight, and if we limit that space to a distance where there are obstacles that participate in our movement decision, then it is possible to claim that we perceive this part of space linearly. Let this space be called "control space". For a human, this space could extend over one hundred meters around him. The obstacle distance estimation in the control space depends on the distance itself. A larger distance will result in larger distance estimation error. The whole control space is divided into two parts: the first one described by the word "ahead", and the second one described by the word "behind". These two categories are relative and they describe our position with respect to objects in the world, therefore they depend on our position and orientation. The model of the control space that we describe with the word "ahead" is well known to us because the building of this space has been supported by the vision system. That part of the control space model can be filled in to tiny details. The control space model that we describe with the word "behind" (compared with "ahead") is always less precise, and depends on our ability to keep in mind details. It also depends on how often "behind" was "ahead" and how long it has been since the last time it happened. The control space building depends on the number of objects/obstacles and their behavior. We experience the "ahead control space" as very intensive and dynamic. The number of obstacles that we take into account of depends on the reqiured level of details. More details less obstacles that we pay attention to. This is achieved by changing the concentration level. The upper limit in the number of obstacles that we attend to mostly depends on the kinematical behavior of obstacles. If there appears only one movable obstacle, the total number of obstacles attended to rapidly drops. The faster the obstacles are, the more complex paths they realize are (in paths we always look for some regularity), and the more movable obstacles there are, the model of control space is poorer. At the moment when speeds are high, paths are very complex and there is a great number of obstacles we are not able to define the world that surrounds us, because it is changing too fast. It is because our capacity of data processing (i.e. the number of active agents) is limited and can not be extended (by age it even drops). We must decide whether we shall use our informatics capacity on details or on an integral picture of the world. The dynamic of "behind control space" is less available for us, and it is based on our extrapolation ability. When our extrapolation ability reaches some limit (which depends on time and the "behind" complexity situation), we turn around (although for a short time), and make that "behind" becomes "ahead" in order to correct extrapolation. Although there are many obvious imperfections of "behind control space", it has some advantages. One of the most significant advantage is that this concept can distinguish important information from less important information. It means that details which we can not comprehend with vision system can not avert our attention; therefore we have more time to attend to important things. When we identify objects/obstacles in the control space, each object obtains a set of properties that describes its relationship to us and to other objects/obstacles. These properties are the size of the object, the estimation of speed and movement direction that are a base for the estimation of collision risk. More precise relation to an object is estimated according to recognized details on it. If we must go around an obstacle, then we estimate the approach, the left and right pass around the obstacle. All the time, the goal of movement is estimated. All these properties have not fixed and precise numerical values. Therefore, the distance to an obstacle can be estimated by a set of values: contact, very close, close, medium, far, very far. The distinction between these six terms is not so precise (except contact) and depends on many factors. One of them is certainly the physical dimension of a person (robot) who (which) estimates the distance property.         Figure 2. Distance and direction perception When we identify objects/obstacles in the control space, each object obtains a set of properties that describes its relationship to us and to other objects/obstacles. These properties are the size of the object, the estimation of speed and movement direction that are a base for the estimation of collision risk. More precise relation to an object is estimated according to recognized details on it. If we must go around an obstacle, then we estimate the approach, the left and right pass around the obstacle. All the time, the goal of movement is estimated. All these properties have not fixed and precise numerical values. Therefore, the distance to an obstacle can be estimated by a set of values: contact, very close, close, medium, far, very far. The distinction between these six terms is not so precise (except contact) and depends on many factors. One of them is certainly the physical dimension of a person (robot) who (which) estimates the distance property. Human direction perception is divided, as mentioned previously, into ahead and behind. The control space ahead could be further divided into: front, left, through left, right and trough right. The control space behind can roughly be divided into only three elements: back, back left and back right. There are two types of object/obstacle size perception. First we can estimate the size of an object as we see it, regardless of the distance of the object. This size is variable and depends on the real distance to the object and the point of view. Beside that, we are trying to estimate the real size of the object, taking into account its estimated distance. The size property can be described by the following set: very small, small, medium big, big, very big and surrounds us. Perception of details on an object is very important for the definition of the relation to the object. As the real distance to an object decreases, the recognition of details on the object becomes more efficient. The recognition of details is very important in the object class recognition and in the individual object recognition. For movement perception it is not enough to have the world model at one moment. We need to have some changes, i.e. we must have the world model of at least two different moments. It means that we must keep in mind time changes of the control space model; therefore, more data processing is necessary, which means more processing time (i.e. intellectual effort). A human has not direct sensors for speed measuring. He estimates the speed indirectly. From two states of the control space, which are separated by the known time interval, the change in distance is estimated and this is a measure for speed estimation. The time interval between two control space states must be short enough to make an illusion of a continuous process. The change of object distance (position) behaves as previously mentioned, and mostly depends on the real distance of objects. That is why the speed estimation of closer objects will be more accurate. In addition, when we estimate the speed of an object, our velocity must be taken into account. It means that we estimate two speeds: the object speed with respect to us and the object speed with respect to the world (i.e. to other objects). From the above, we can conclude that the speed estimation is a very demanding process (some creatures with simple neural system developed a different approach, simpler and less efficient). That is why it is very hard to estimate the speed of more than one object at a time, and why the acceleration estimation is almost impossible. If we have to estimate acceleration then we try to use some different information that is not of visual type (a sound, for example). The speed property can be defined by the following set: rest, very slow, slow, medium fast, fast and very fast; and the speed direction property by the set: toward us, toward us right, toward us left, from us, from us right and from us left. The perception of the risk of collision with some obstacle is based on the estimation of the distance of the obstacle, on its speed perception and on the known speed of the observer. If some obstacle is in motion it will be very useful if we know its predefined path. It is reasonabereasonable to presume [10] at least two degrees of collision risk: static and kinematical. The static degree of collision risk is defined by the estimated distance of the obstacle, while the kinematical degree of collision risk is defined by the estimated speed of the obstacle (amount and direction). The total degree of collision risk is a combination of previously mentioned factors. The collision risk estimation is very important in the movement decision making, and could be defined by the set: no risk, small risk, medium risk, great risk and very great risk. Table 1. Properties of obstacles in the control space Identification properties of obstaclesattachment classattachment subclass 1attachment subclass 2attachment subclass 3attachment subclass npersonal identification1 2 3 4 5 6 7house car tree wall man fountain . . .male female baby child girl mid age old age very old blonde black brunette Jo Ann Maria . . . Identification properties of obstaclesdistancedirectionsizevisibilityvisible detailposition in view field1 2 3 4 5 6 7 8contact very close close medium far very farfront left through left right through right back back left back rightvery small small medium big big very big surrounds usall left side right side central part by parts not visiblefront part back part left side right sideon left edge on right edge on l. and r. edge not on edge Identification properties of obstaclesobstacle betweenspeedspeed directioncollision dangerleft going aroundright going around1 2 3 4 5 6 7(obstacle designation)rest very slow slow medium fast fast very fasttoward us tow. us right tow. us left from us from us right from us leftno danger small danger med. danger great danger v. gr. dangerno pass v. nar. pass narrow pass medium pass wide pass open passno pass v. nar. pass narrow pass medium pass wide pass open pass Table 1 Properties of obstacles in the control space The perception of approach to some object is connected with the movement goal. The approach perception is closely related with the recognition of details on some object. Perception of going around is important if we intend to go around an obstacle. This perception depends on the relationship between objects and on the size of the one that wants to go around. Going around left and right property can be defined by the set: no pass, very narrow pass, narrow pass, medium pass, wide pass and open pass. Visibility perception of an object directly depends on results of picture analysis. It is a complex function of the space distribution of objects, their size and shape, but it also depends on our relation (position and orientation) to other objects. The visibility perception could be defined by the set: all, left side, right side, central part, by parts and not visible. The visibility property is connected to another property which tells us about the object position in relation to the view field. This property gives us the information on whether some part of the object crosses the view field limits. If that is the case, then the visibility estimation and some other estimations (distance, size, etc.) are not comprehensive because they are based on incomplete information. In that case we turn around to put a complete object in the view field. The property of the position in the view field could be defined by the set: on left edge, on right edge, on left and right edge, not on edge. If some object is not completely visible, we are also interested in which object covers it, i.e. which object stands between the attended to object and us. Goal movement perception could be one of the most abstract ideas. It can be phrased by a set of conditions which are in complex relations. Each of these conditions and relations may be interpreted very freely; therefore, a definition of a final goal can become very fuzzy and can have more meanings. In that case we can talk only about the probability of reaching the goal. From the other point of view, the goal description could be very simple and material; therefore the reaching of goal is trivial. General situation perception is the estimation of the global situation at some moment. It does not refer to any particular obstacle or object, it refers to the complete relation of all objects (relation between object themselves and objects to human). The general situation estimation depends on the properties of all objects in the control space, but it is not merely the sum of all properties. The general situation is not a very dynamic value and it changes slower than the properties of individual obstacles. 4. CONCLUSION For the robot path planning using the methods of cognitive robotics, an adequate space model needs to be built. Instead of a quantified and comprehensive mathematical model, an adequate mental model of space is built. It consists of a qualitative and estimated description of obstacle identification properties and their relationships. The memory size required for that kind of model is significantly reduced when compared with a classic model (configuration space, for example), but elements of the mental space model are complex categories obtained by the processing of estimated data. This data processing is carried out by agents (programs) specialized for a corresponding object property (i.e. two objects start to cover each other, one object exits from the view field etc.). It is necessary to find an adequate set of object properties which permits a robot movement that is comparable with a human movement. Since the number of agents that can be active at some moment is limited, it is necessary to find the right order for (de)activation of agents. This will permit an efficient and natural robot movement by relatively small model of world space. 5. LITERATURE K. Kawamura;, et all:, A Parallel Distributed Cognitive Control System for a Humanoid Robot, International Journal of Humanoid Robotics, Vol. 1, No. 1 (2004), pp 65-93 S. Kasderidis;, J.G. Taylor:, Attention-based Learning, Proceedings from International Joint Conference on Neural Networks and Fuzzy Systems, Budapest, Hungary, 25-29 July 2004 J.G. Taylor;, N. Fragopanagos:, Modeling Human Attention and Emotions, Proceedings from International Joint Conference on Neural Networks and Fuzzy Systems, Budapest, Hungary, 25-29 July 2004 X. Huang;, J. Weng:, Value Systems Development for a Robot, Proceedings from International Joint Conference on Neural Networks and Fuzzy Systems, Budapest, Hungary, 25-29 July 2004 S. Wood:, Representation and purposeful autonomous agents, Robotics and Autonomous Systems 49 (2004), 79-90 T. Wagner;, U. Visser;, O. Herzog:, Egocentric qualitative spatial knowledge representation for physical robots, Robotics and Autonomous Systems 49 (2004), 25-42 S.B. Badia;, P. Verschure:, A Collision Avoidance Model Based on the Lobula Giant Movement Detector (LGMD) Neuron of the Locust, Proceedings from International Joint Conference on Neural Networks and Fuzzy Systems, Budapest, Hungary, 25-29 July 2004 T. Barbera;, et all:, How task analysis can be used to derive and organize the knowledge for the control of autonomous vehicles, Robotics and Autonomous Systems 49 (2004), 67-78 N.L. Cassimatis;, at all:, Integration cognition, perception and action through mental simulation in robots, Robotics and Autonomous Systems 49 (2004), 13-23 T. Kubota;, H. Hashimoto:, A Strategy for Collision Avoidance Among Moving Obstacle for a Mobile Robots, 11th IFAC World Congress, Volume 9 Tallinn 1990, 103-108 This research is supported by the Croatian Ministry of Science, Education and Sport, Project No. 0120002/2002. CIM05-eng.doc  PAGE 10 M. Crnekovi, M. Su evi, D. Brezak, J. Kasa Cognitive robotics and robot path planning  PAGE 11 Lumbarda, Kor ula, 2005 10th INTERNATIONAL SCIENTIFIC CONFERENCE ON PRODUCTION ENGINEERING CIM 2005  Published by: R. Cebalo$(*02XZ\b|~ ɸj[O[ChwCJOJQJ^JhgaCJOJQJ^Jh:h$CJOJQJ^J3hiGgh+6CJOJQJ^JaJF*63hiGgh$6CJOJQJ^JaJF*63hiGghx6CJOJQJ^JaJF*6 h:hxCJOJQJ^JaJ#h:hI5CJOJQJ^JaJ#h:h+5CJOJQJ^JaJ#h:h$5CJOJQJ^JaJZ\~t  & ( %$ d^`a$gdoM&d&^`$d`a$gdJ% $da$gdJ%dgdJ% $da$gdJ%Vy * ( 8 < > L N \ ` d p  ǨǨǨǙxfxUD h:hX$CJOJQJ^JaJ h:h4CJOJQJ^JaJ"h:h+5CJOJQJ\^J"h:hv5CJOJQJ\^Jh:h$CJOJQJ^Jh:h=CJOJQJ^Jh:hvCJOJQJ^Jh:h+5CJOJQJ^Jh:h+CJOJQJ^JhgaCJOJQJ^Jh:h$CJOJQJ^Jh:hwCJOJQJ^J       * , 3 ; F K L O W e p w ӴŴŴpp_N h|h|CJOJQJ^JaJ h|h_`CJOJQJ^JaJ h|hgaCJOJQJ^JaJ h|h@-CJOJQJ^JaJ h|hhbCJOJQJ^JaJ h:h@-CJOJQJ^JaJ h:hhbCJOJQJ^JaJh|CJOJQJ^JaJ h:hX$CJOJQJ^JaJhwCJOJQJ^JaJhgaCJOJQJ^JaJ       % - 1 : Z [ n s  :BC[^frͿͿ͝{jjj\j\hyCJOJQJ^JaJ h:hOD1CJOJQJ^JaJ h:hzICJOJQJ^JaJ h:hfCJOJQJ^JaJ h:h/$CJOJQJ^JaJ h:h@-CJOJQJ^JaJh|CJOJQJ^JaJ h:hfSTCJOJQJ^JaJ h|h|CJOJQJ^JaJ h|hfSTCJOJQJ^JaJ$rаteSA0 h:h_HCJOJQJ^JaJ"h:hJ%5CJOJQJ\^J"h:h)+5CJOJQJ\^Jhy5CJOJQJ\^J;hyhef5CJOJQJ\^JcHdhdhdhJ&;hyhef5CJOJQJ\^JcHdhdhdhK&?h:hJ%h?WFCJOJQJ^JaJcHdhdhdhN& h:hX$CJOJQJ^JaJhyCJOJQJ^JaJ h:hOD1CJOJQJ^JaJ ~t$ h@da$gdJ%$d`a$gdJ%$@Efgns/0DHPZ\`koy޼޼ޫޫ{m\\ h:hdCJOJQJ^JaJhRpCJOJQJ^JaJ h:hwCJOJQJ^JaJh!NCJOJQJ^JaJ h:hdCJOJQJ^JaJ h:hCJOJQJ^JaJ h!NhCJOJQJ^JaJ h!NhqSCJOJQJ^JaJ h!NhCJOJQJ^JaJ h!Nh)+CJOJQJ^JaJ$y,-8:>JMgпᮿпᝌrXr3 *h ZhRpCJOJQJ^JaJ\&* *3 *h Zh=CJOJQJ^JaJ\&* * h:h?8CJOJQJ^JaJ h:h=CJOJQJ^JaJ h:h!NCJOJQJ^JaJ h:hCk`CJOJQJ^JaJ h:hdCJOJQJ^JaJh!NCJOJQJ^JaJ h:hdCJOJQJ^JaJ*9пcRD6R(R(hGCJOJQJ^JaJh sPCJOJQJ^JaJhwCJOJQJ^JaJ h:h 7CJOJQJ^JaJ?h:hxCh?WFCJOJQJ^JaJcHdhdhdhO&9hxCh?WFCJOJQJ^JaJcHdhdhdhO&h)+CJOJQJ^JaJ h:hddCJOJQJ^JaJ h:h=CJOJQJ^JaJ h:h?8CJOJQJ^JaJ h:h=CJOJQJ^JaJhxCCJOJQJ^JaJ9;=GKVW%HLTf,6;<=Aӱӱӱӱӱӱӱshj[CJOJQJ^JaJ h:hCCJOJQJ^JaJ h:hKCJOJQJ^JaJhKCJOJQJ^JaJ h:he~CJOJQJ^JaJ h:hCJOJQJ^JaJh iCJOJQJ^JaJhGCJOJQJ^JaJ h:h 7CJOJQJ^JaJ'ABCGdmxyѷ|R8!-h;Jh?8CJOJQJ^JaJl&*3 *h;Jh?8CJOJQJ^JaJl&* *R *h;Jhj[h;JCJOJQJ^JaJcHdhdhdhl&l&* *@Hhl& *h;Jh;JCJOJQJ^JaJl&* *3 *h;Jhj[CJOJQJ^JaJl&* *3 *h;JhCCJOJQJ^JaJl&* *-h;JhCCJOJQJ^JaJl&*-h;Jhe~CJOJQJ^JaJl&* '@HLPaluw}~Ϳͱͱͣͣ͒uXG h:h vCJOJQJ^JaJ9h vh?WFCJOJQJ^JaJcHdhdhdhO&9hh?WFCJOJQJ^JaJcHdhdhdhO& h:h CJOJQJ^JaJh vCJOJQJ^JaJhj[CJOJQJ^JaJh sPCJOJQJ^JaJ h:hFCJOJQJ^JaJ h:hCCJOJQJ^JaJ h:h?8CJOJQJ^JaJ*/JNcfstxz{:=п~p~~_p_p_Np h:h/CJOJQJ^JaJ h:h?8CJOJQJ^JaJh/CJOJQJ^JaJ h:h bCJOJQJ^JaJ h:hiCJOJQJ^JaJ h:hCJOJQJ^JaJhaCJOJQJ^JaJ h:h9CJOJQJ^JaJh9CJOJQJ^JaJ h9h9CJOJQJ^JaJ h:hvCJOJQJ^JaJߵziz[Jz[zJ9([ h:hwCJOJQJ^JaJ h:hvCJOJQJ^JaJ h:h,CJOJQJ^JaJh/CJOJQJ^JaJ h:h bCJOJQJ^JaJ h:hiCJOJQJ^JaJR *h Zh/h ZCJOJQJ^JaJcHdhdhdh]&]&* *R *h Zh bh ZCJOJQJ^JaJcHdhdhdh]&]&* *@Hh]& *h Zh ZCJOJQJ^JaJ]&* * "&6<EFJNX_ehqrtx  *+-ABCDKRUY\cfjklmw h:h?8CJOJQJ^JaJhCJOJQJ^JaJ h:h bCJOJQJ^JaJh/CJOJQJ^JaJ h:hwCJOJQJ^JaJ=%+߿xfxTC2! hh\CJOJQJ^JaJ hhWCJOJQJ^JaJ hh$CJOJQJ^JaJ"hh$5CJOJQJ\^J"hh\5CJOJQJ\^J"hh b5CJOJQJ\^J *h5CJOJQJ\^J *h> *hh?WF5CJOJQJ\^JcHdhdhdhO&?h:hxh?WFCJOJQJ^JaJcHdhdhdhO&?h:hwh?WFCJOJQJ^JaJcHdhdhdhO& $!"D$%%%$ da$gd$d`gd?WFoO&d&F$ h@da$ d`gd$dgdJ%+(,GJ`ad ;<?@RVcgoxͼ~m~~~\~\~\~\~ h:h0CJOJQJ^JaJ h:hh@CJOJQJ^JaJh cCJOJQJ^JaJ h:hBUCJOJQJ^JaJ h:hdHCJOJQJ^JaJhCJOJQJ^JaJ h:hZCJOJQJ^JaJ h:hWCJOJQJ^JaJ hh\CJOJQJ^JaJ hhWCJOJQJ^JaJ!x|"+3FJVWrx Ϳޠ޿ޱޏ~~~~~~~~~~ h:hCJOJQJ^JaJ h:h|_CJOJQJ^JaJ h:hZCJOJQJ^JaJh cCJOJQJ^JaJhdrCJOJQJ^JaJ h:hdrCJOJQJ^JaJ h:h0CJOJQJ^JaJ h:h cCJOJQJ^JaJ1 047:=>AMN\biq!*+,8EGOg᠏~m\~ h:hdrCJOJQJ^JaJ hdrhdrCJOJQJ^JaJ h:hCJOJQJ^JaJ h:hUCJOJQJ^JaJ h:hWCJOJQJ^JaJ h:hh@CJOJQJ^JaJhdrCJOJQJ^JaJ h:h|_CJOJQJ^JaJ h:hZCJOJQJ^JaJh$CJOJQJ^JaJ$gk~     & ) ? @ E H J c p q u пЮН~~~m~ h:hCJOJQJ^JaJ h:h `CJOJQJ^JaJhNfCJOJQJ^JaJ h:hn}CJOJQJ^JaJ h:hCJOJQJ^JaJ h:hq1CJOJQJ^JaJ h:hCJOJQJ^JaJ h:hUCJOJQJ^JaJhCJOJQJ^JaJ* !!#!$!+!.!!!!!!!"""("*"i"z"}"!#-#^#п~~mm\m h:ho[CJOJQJ^JaJ h:hrlvCJOJQJ^JaJ h:h*CJOJQJ^JaJ h:hUCJOJQJ^JaJ h:hCJOJQJ^JaJh}6CJOJQJ^JaJ h:h*CJOJQJ^JaJ h:hdHCJOJQJ^JaJhNfCJOJQJ^JaJ h:h `CJOJQJ^JaJ^#e#l#s#########$$"$3$7$庩|k|kZIZ8 h:h bsCJOJQJ^JaJ h:hm3CJOJQJ^JaJ h:hzCJOJQJ^JaJ h:hceCJOJQJ^JaJh}6CJOJQJ^JaJhT/CJOJQJ^JaJ h:hT/CJOJQJ^JaJ h:h}6CJOJQJ^JaJ h:hrlvCJOJQJ^JaJ3 *h Zhm3CJOJQJ^JaJ^&* *3 *h ZhrlvCJOJQJ^JaJ^&* *7$9$:$@$B$C$D$N$Q$W$Z$[$$$$$$$$$$$$N%P%Y%]%^%b%r%%%%±䠁p_ h:h*CJOJQJ^JaJ h:ho[CJOJQJ^JaJhT/CJOJQJ^JaJ h:hT/CJOJQJ^JaJ h:h65CJOJQJ^JaJ#h:hQf5CJOJQJ^JaJh bs5CJOJQJ^JaJ#h:h$5CJOJQJ^JaJ h:h$CJOJQJ^JaJ&&&&&&&&&&&&&&&&'<'H'M'[''''''''''''''޼޼Ｋi?h:hK sh?WFCJOJQJ^JaJcHdhdhdhP& h:h#CJOJQJ^JaJ h bshCJOJQJ^JaJ h bsh-CJOJQJ^JaJ h bsh|CJOJQJ^JaJ h bsh\CJOJQJ^JaJ h bshc_CJOJQJ^JaJ h bshCJOJQJ^JaJ!%''*a-c-d-e-f-j-k-n-o-p-q-r-s-v-w-x-y-z-}----- $ ha$gdJ%$ hda$gdJ%'''+(0(N(Q(n(p(s(y((((((( ))))$)+)0):)q)w))))))))****3*=*_*νyyyy h bsh^CJOJQJ^JaJ h bshX$CJOJQJ^JaJ h bsh-CJOJQJ^JaJ h bsh5TCJOJQJ^JaJ h bshK sCJOJQJ^JaJ'HhP&h?WFCJOJQJ^JaJ9h-h?WFCJOJQJ^JaJcHdhdhdhP&(_*j*n****************A+E+L+N++++++ͼxgVEVVVV h bshCJOJQJ^JaJ h bsh[_RCJOJQJ^JaJ h bshO"CJOJQJ^JaJ h bsh=CJOJQJ^JaJ h bshb'RCJOJQJ^JaJ h bsh: CJOJQJ^JaJ h bsh bsCJOJQJ^JaJ h:h: CJOJQJ^JaJ h:h#CJOJQJ^JaJ h bshNCJOJQJ^JaJ h bsh|CJOJQJ^JaJ++,,8,E,I,V,h,r,y,,,,,,-7-;-`-a-b-c-f-h-޼n]B4jh:hCJOJQJU^JaJmHnHu h:h"|]CJOJQJ^JaJ4jh:hhCJOJQJU^JaJmHnHu h:hhCJOJQJ^JaJ h bshO"CJOJQJ^JaJ h bsh3CJOJQJ^JaJ h bsh3CJOJQJ^JaJ h bshCJOJQJ^JaJ h bsh[_RCJOJQJ^JaJ h bshb'RCJOJQJ^JaJh-i-k-m-q-r-s-u-z-{-|-}-~-------ԹԨ|a||ԗH2*Hh3Fh7-6CJOJQJ^JaJ0hiGghO"CJOJQJ^JaJF*64jh:hkCJOJQJU^JaJmHnHu4jh:hCJOJQJU^JaJmHnHu h:h)&CJOJQJ^JaJ h:hrCJOJQJ^JaJ4jh:hhCJOJQJU^JaJmHnHu h:hhCJOJQJ^JaJ4jh:huCJOJQJU^JaJmHnHu--------!0"0#0$03q5.66$ h@da$gdJ%$ da$gd$ndgdJ%$ hda$gdJ%$ hda$gdJ% $ ha$gdJ%-------. . ...1.9.:.x.y.........εqq`O>q>q>`>q>q h bshoCJOJQJ^JaJ h bsh bsCJOJQJ^JaJ h bsh$nCJOJQJ^JaJ h bsh3CJOJQJ^JaJ h bsh3CJOJQJ^JaJ h:h3CJOJQJ^JaJ h:h"|]CJOJQJ^JaJ0hiGgh"|]CJOJQJ^JaJF*60hiGghO"CJOJQJ^JaJF*60hiGghhCJOJQJ^JaJF*6.../////#/U/^/j/n//////00 0!0#0Ｋ͚މ{͚mP9h3h?WFCJOJQJ^JaJcHdhdhdhP&h3CJOJQJ^JaJh CJOJQJ^JaJ h bshCJOJQJ^JaJ h bsh7FCJOJQJ^JaJ h bsh$nCJOJQJ^JaJ h bsh4CJOJQJ^JaJ h bsh3CJOJQJ^JaJ h bsh bsCJOJQJ^JaJ h bshoCJOJQJ^JaJ#0$0%0*0U0[0c0e0r0t00000s1u1{1|111111111122E2νyhhWyWWyWW hhL(CJOJQJ^JaJ hhR CJOJQJ^JaJ hhIdCJOJQJ^JaJ hhCJOJQJ^JaJ hhKNCJOJQJ^JaJ hhao5CJOJQJ^JaJ hh@zCJOJQJ^JaJ h:hao5CJOJQJ^JaJ?h:h3h?WFCJOJQJ^JaJcHdhdhdhP&E2F2L2f2i2u222222O3T33333333333333޹͆x͆xxgVE h:hIdCJOJQJ^JaJ h:h3CJOJQJ^JaJ h:h-CJOJQJ^JaJhCJOJQJ^JaJ hh-CJOJQJ^JaJ hh@zCJOJQJ^JaJ hhCJOJQJ^JaJ&hh bsh bsCJOJQJ^JaJ hhIdCJOJQJ^JaJ hh bsCJOJQJ^JaJ hhCJOJQJ^JaJ3 4<4J4]4^4r4z4444444444444 55<5A5C5J5_5b5p5q5r5u5555xggg hM5hjDCJOJQJ^JaJ hM5hao5CJOJQJ^JaJ hM5hvCJOJQJ^JaJ hM5h2CJOJQJ^JaJ hM5h%CJOJQJ^JaJ hM5hCJOJQJ^JaJ hM5hM5CJOJQJ^JaJ hM5hiCJOJQJ^JaJ hM5h7CJOJQJ^JaJ"5566#6,6-6.6/666?6D6H6T6W6n6o6666666޼ޫ{j{\K:: h:h!CJOJQJ^JaJ h:h>j_CJOJQJ^JaJhCJOJQJ^JaJ h:hM5CJOJQJ^JaJhM5CJOJQJ^JaJ h:h_CJOJQJ^JaJ h:hjCJOJQJ^JaJ h:hIdCJOJQJ^JaJ hM5hCJOJQJ^JaJ hM5hCJOJQJ^JaJ hM5hjDCJOJQJ^JaJ hM5h>j_CJOJQJ^JaJ666MX$ & F h@dEƀJ&^a$gdJ%X$ & F h@dEƀJ&^a$gdJ%666MX$ & F h@dEƀJ&^a$gdJ%X$ & F h@dEƀJ&^a$gdJ%6666666666666666667 7-7/7ònZFn8hM5CJOJQJ^JaJ&hM5hX&.56CJOJQJ^JaJ&hM5hu56CJOJQJ^JaJ hM5huCJOJQJ^JaJ h:h CJOJQJ^JaJ hM5hM5CJOJQJ^JaJ hM5h!CJOJQJ^JaJ hM5h_CJOJQJ^JaJ h:h!CJOJQJ^JaJ h:h_CJOJQJ^JaJ3 *h ZhM5CJOJQJ^JaJ^&* *666MX$ & F h@dEƀJ&^a$gdJ%X$ & F h@dEƀJ&^a$gdJ%666MX$ & F h@dEƀJ&^a$gdJ%X$ & F h@dEƀJ&^a$gdJ%6669;==>>GP$ h@dEƀ䥓a$gdJ%$ h@da$gdJ%X$ & F h@dEƀJ& ^a$gdJ%/7?7B7L7U7m7777777778888J8M8Q8T8e8l888888888899 993989\9b9g9n9s9z9999ӠӠӠӠӠ~ h:h`#CJOJQJ^JaJ h:huCJOJQJ^JaJ hM5h1ZuCJOJQJ^JaJ hM5hX&.CJOJQJ^JaJ hM5hLCJOJQJ^JaJhv#CJOJQJ^JaJhM5CJOJQJ^JaJ hM5huCJOJQJ^JaJ/99999::.:9:f:g:v:~::::::;;;;6;:;S;W;h;k;;;;ͼͫͼ͚{{{{jY h:h`BCJOJQJ^JaJ h:huCJOJQJ^JaJ hzhZ^CJOJQJ^JaJhzCJOJQJ^JaJ hzhxCJOJQJ^JaJ hzh`BCJOJQJ^JaJ hzhLCJOJQJ^JaJ hzh`#CJOJQJ^JaJ hG~hLCJOJQJ^JaJ hG~h`#CJOJQJ^JaJ;;;;;;;;;;;;;<<<<'<(<)<+<,</<0<=<`<<<<<<<<<<==>=B=K=Y=c=l======ͼ͚{{hF2CJOJQJ^JaJ hF2hHCJOJQJ^JaJ hF2hQCJOJQJ^JaJ hF2hCJOJQJ^JaJ hF2hCJOJQJ^JaJ hF2h`BCJOJQJ^JaJ hF2hF2CJOJQJ^JaJ hF2hw1CJOJQJ^JaJ/========= > >">#>&>->.>3>:>I>J>W>а}l}[}[}lG}6}l} hF2h1CJOJQJ^JaJ&hF2hhCJOJQJ^JaJ hF2hMACJOJQJ^JaJ hF2hCJOJQJ^JaJ hF2hCJOJQJ^JaJ hF2h+\CJOJQJ^JaJ h:hCJOJQJ^JaJ?h:hw1h?WFCJOJQJ^JaJcHdhdhdhQ&hw1CJOJQJ^JaJ h:hCJOJQJ^JaJ hF2h=CJOJQJ^JaJW>g>m>s>w>x>>>>>>>>>>>>>?ͼͫͫ͌oO>0>hI5/CJOJQJ^JaJ h:hX'CJOJQJ^JaJ?h:hI5/h?WFCJOJQJ^JaJcHdhdhdhQ&9hI5/h?WFCJOJQJ^JaJcHdhdhdhQ&h`#CJOJQJ^JaJ hF2hCJOJQJ^JaJ hF2h=CJOJQJ^JaJ hF2h1CJOJQJ^JaJ hF2hMACJOJQJ^JaJ hF2hCJOJQJ^JaJ hF2h<CJOJQJ^JaJ>>>>>>>>>>>>>>>>>CFXJZJ\J]J`JbJcJdJ $ h@a$gdJ%$ h@da$gdJ%???#?I?V?Z?q?t???????пbH3 *h Zh10CJOJQJ^JaJ_&* *@Hh_& *h Zh ZCJOJQJ^JaJ_&* *3 *h ZhX'CJOJQJ^JaJ_&* * h10hCJOJQJ^JaJ h10h10CJOJQJ^JaJ h10hX'CJOJQJ^JaJ h:hX'CJOJQJ^JaJ h:hCJOJQJ^JaJhCJOJQJ^JaJ????? @@@@"@%@&@;@D@]@b@f@s@@@@@@@@ŴŴŴŴssbssQ h:h sPCJOJQJ^JaJ h:h_h^CJOJQJ^JaJcHdhdhdhY&?h:h;h^CJOJQJ^JaJcHdhdhdhY&YJZJ[J]J^J_J`JaJiJjJkJvJwJxJyJzJJJJJJJèÍ||||kR90hiGgh0CJOJQJ^JaJF*60hiGghOCJOJQJ^JaJF*6 h:h%NCJOJQJ^JaJ h:h>CJOJQJ^JaJ4jh:h[_BCJOJQJU^JaJmHnHu4jh:hUF(CJOJQJU^JaJmHnHu h:h0CJOJQJ^JaJ4jh:h%NCJOJQJU^JaJmHnHu h:hoYCJOJQJ^JaJdJeJfJgJhJiJlJmJnJoJpJqJrJsJtJuJvJwJyJ{J|J}J~JJJJ$ h@da$gdJ% $ h@a$gdJ%JJJJJJJJJkKKKKKKKKKKLL;L>LULXL`LaLfLwLLLLLLLLL*MѸыt`t`t`t`t`t`t`t`t`t`t`t`t`t`t`t'HhY&h^CJOJQJ^JaJ-HhY&h:h^CJOJQJ^JaJ'HhY&hmCJOJQJ^JaJ0hiGgh=CJOJQJ^JaJF*60hiGghOCJOJQJ^JaJF*60hiGgh0CJOJQJ^JaJF*6*Hh3Fh7-6CJOJQJ^JaJ%JJJyNzNOQRfW7Y)Z{]$ h@da$gdJ%Q$ h@dC$EƀY&a$gd^$ h@da$gdy6 *M+M,M0MTmTnToTpTrTTTTTTTTTTTTT7U8U=UyUUUUUUUUUUUUU VVV"V򿮿Н򌝌~~Ќ~ЌЌ~~hO@CJOJQJ^JaJ h:hYCJOJQJ^JaJ h:h_JCJOJQJ^JaJ h:h~CJOJQJ^JaJ h:hbCJOJQJ^JaJ h:hGUCJOJQJ^JaJ h:h7CJOJQJ^JaJh!mCJOJQJ^JaJ-"V)V*V7V;VHVKVtVVVVVVVVVV WW"W&W3W@WCWGWVWcWeWfWkWvWWWWWWWWWͿͿͿͿͿͿͿͿͿͮ͌͝{m{m{mmh4CJOJQJ^JaJ h:hCJOJQJ^JaJ h:hGCJOJQJ^JaJ h:hGUCJOJQJ^JaJ h:h~CJOJQJ^JaJhO@CJOJQJ^JaJ h:hC"CJOJQJ^JaJ h:hYCJOJQJ^JaJ h:hO@CJOJQJ^JaJ&WWWWWXX X X!X/X4XKXOXXXXXXXX Y Y'Y(Y6Y7Y t0644 la{_}______________________ ````(`$ h@$Ifa$gdJ%_____________________________ ` ````````$`'`(`0`1`4`5`8`9`۷۷۷ۨۄu۷h 6CJOJQJ^JaJ#h:h!6CJOJQJ^JaJ#h:h9+6CJOJQJ^JaJh9+6CJOJQJ^JaJ#h:hW4 6CJOJQJ^JaJ#h:hJ6CJOJQJ^JaJ#h:h`E6CJOJQJ^JaJ#h:hd6CJOJQJ^JaJ.(`5`9`C`N`[`d`p`{````````$ h@$Ifa$gdJ%9`:`;`B`C`M`N`Z`[`c`d`o`p`z`{````````````````````̩vd#h:h`E5CJOJQJ^JaJ h:hQCJOJQJ^JaJ h&.h"CJOJQJ^JaJ h:hB8CJOJQJ^JaJ h:h`ECJOJQJ^JaJ#h:hy66CJOJQJ^JaJ#h:h`E6CJOJQJ^JaJh 6CJOJQJ^JaJ#h:hW4 6CJOJQJ^JaJ!``!kd$$Ifl֞f PY $?> t062 944 la`````aa%a6aAaHaTa[asgkd$$Ifl $& t0644 la$ h@$Ifa$gdJ%$ h@a$gdJ% `aaaa$a%a5a6a:a@aAaMaSaTaZa[a\ajakaaaaaaaaaaaaaaaaۺsasasasasasasa#h:h`E6CJOJQJ^JaJ#h:hKjP6CJOJQJ^JaJ#h:h~6CJOJQJ^JaJ h:h`ECJOJQJ^JaJ#h:h }5CJOJQJ^JaJ#h:h!5CJOJQJ^JaJhR~85CJOJQJ^JaJ#h:h`E5CJOJQJ^JaJ#h:hv5CJOJQJ^JaJ"[a\a%kd$$Ifl֞x j,a $B t0644 la\a^a`abadafahajaaaaaaaaaaaaaabbb)b7b?bLb$ h@$Ifa$gdJ%aaaaaaaaaaaaaa䥌fC$=Hhc&hwChwC6CJOJQJ^JaJc&*EHhb&hwChwC6CJOJQJ^JaJc&*CJaJKHhb& *hwChwC6CJOJQJ^JaJc&* *CJaJ0hwChKjP6CJOJQJ^JaJc&*0hwCh`E6CJOJQJ^JaJc&*KHhb& *hwCh Z6CJOJQJ^JaJc&* *CJaJ6 *hwChKjP6CJOJQJ^JaJc&* * aaaaaaabbbbbb(b)b6b7b>b?bKbLbWbXbcbdbmbnbwbxbbbb缪tttttttb#h:h }6CJOJQJ^JaJ#h:h }6CJOJQJ^JaJ#h:hW|6CJOJQJ^JaJ#h:h`E6CJOJQJ^JaJ#h:h1kz6CJOJQJ^JaJ#h:h`E6CJOJQJ^JaJ0hwCh`E6CJOJQJ^JaJc&*0hwChKjP6CJOJQJ^JaJc&*LbXbdbnbxbbbbbbb$ h@$Ifa$gd }$ h@$Ifa$gdJ% bbkd$$Ifl֞x j,a $B t062 92 92 944 labbbbcdii`kambmpmpqqr$ h@da$gd F$ h@da$gd& $ h@da$gdJ%bbbbbbbbbc c c.c2c߿|kWkCk2$h%%CJOJQJ^JaJ h:hCJOJQJ^JaJ&h:h%%56CJOJQJ^JaJ&h:h56CJOJQJ^JaJ h%%56CJOJQJ^JaJEh:h hiGg56CJOJQJ^JaJcHdhdhdhF?h:h^hiGgCJOJQJ^JaJcHdhdhdhF?h:h"hiGgCJOJQJ^JaJcHdhdhdhF?h:h9hiGgCJOJQJ^JaJcHdhdhdhF 2cAcFcaccciclcqcuccccccccccccccd d.d7d:d>dBdGdKdLd]d`ddddＨuuuuuuguuhR~8CJOJQJ^JaJ h:hCJOJQJ^JaJ h:hcGCJOJQJ^JaJ hR~856CJOJQJ^JaJ&h:h56CJOJQJ^JaJ&h:h56CJOJQJ^JaJ h:h%%CJOJQJ^JaJh%%CJOJQJ^JaJ h:hCJOJQJ^JaJ$dddddddeeYe]eoeseze~eeeeeeeeeeeefffLfQf\f]f˷veeeveeeSeSe#h:hLbK6CJOJQJ^JaJ h:hLbKCJOJQJ^JaJ h:h(CJOJQJ^JaJ h:h%%CJOJQJ^JaJh%%CJOJQJ^JaJ h:hXCJOJQJ^JaJ&h:hX56CJOJQJ^JaJ h:hCJOJQJ^JaJ h:hCJOJQJ^JaJ#h:h6CJOJQJ^JaJ ]f^fcfffffffffffffggg"g.g0g1g5gKg]gagbgfgggggggggggggggghhh#h.h2h>hBhJhlh~hhhhббР РРЎ#h:h(6CJOJQJ^JaJ h:h!{CJOJQJ^JaJ h:hoCJOJQJ^JaJhdCJOJQJ^JaJ h:h(CJOJQJ^JaJh%%CJOJQJ^JaJ h:hYjCJOJQJ^JaJ8hhhh ii.i;iJiNiViYihiiiiiiiiiiiijj j jjj-j.j޿xxjjjjY hJ3hJ3CJOJQJ^JaJhJ3CJOJQJ^JaJ h:hR7CJOJQJ^JaJ h:hCJOJQJ^JaJ&h:h>56CJOJQJ^JaJ h:hcGCJOJQJ^JaJhdCJOJQJ^JaJ h:h!{CJOJQJ^JaJ h:h(CJOJQJ^JaJ.j9jDjEj|jjjjjjjjjjjjjjk?kCkKkTkVk_k`k|kkkkkkkk޿ޮЮЮЮ~jYYYY h:hFCJOJQJ^JaJ&h:hF56CJOJQJ^JaJ h:hcGCJOJQJ^JaJ h:hlCJOJQJ^JaJhlCJOJQJ^JaJ h:hR7CJOJQJ^JaJ h:hGCJOJQJ^JaJhJ3CJOJQJ^JaJ h:hCJOJQJ^JaJhlCJOJQJ^JaJ h:hFCJOJQJ^JaJ h:hoCJOJQJ^JaJ"mmmmmm nn"nAnDnHnInNnVnZn\ngnnnnnnnnnnno oo&o.oAoNoaoeooooooooooppipmp~pppp±Рбб±±±± h:h3&!CJOJQJ^JaJ h:hoCJOJQJ^JaJ h:hCJOJQJ^JaJhCJOJQJ^JaJ h:h]UCJOJQJ^JaJ h:h:XCJOJQJ^JaJhlCJOJQJ^JaJ5pppq q4q7qgqkqqqqqqqqqqrrrr￮x]I'Hh"Fh.FCJOJQJ^JaJ5h.Fh\cCJOJQJ^JaJ F*CJaJ#h:hS5CJOJQJ^JaJ#h:h\c5CJOJQJ^JaJ#h:h,5CJOJQJ^JaJ h:hoCJOJQJ^JaJ h:hCJOJQJ^JaJ h:hCJOJQJ^JaJhCJOJQJ^JaJ h:hO#CJOJQJ^JaJrrrrrrrbrrrպ`B'5h.FhCJOJQJ^JaJ F*CJaJ;h.Fhm9@6CJOJQJ^JaJ F*6CJaJTh.Fhm9@h.FCJOJQJ^JaJcHdhdhdh!F F*CJaJ'Hh!Fh.FCJOJQJ^JaJ5h.Fhm9@CJOJQJ^JaJ F*CJaJ5h.Fh\cCJOJQJ^JaJ F*CJaJTh.Fh\ch.FCJOJQJ^JaJcHdhdhdh"F F*CJaJ rr`sMX$ & F @dEƀJ&.^a$gdX$ & F @dEƀJ&.^a$gdrrrrrrrrkslsms}s~sssssfHf*H;h.Fh:6CJOJQJ^JaJ F*6CJaJ;h.Fh6CJOJQJ^JaJ F*6CJaJTh.Fhh.FCJOJQJ^JaJcHdhdhdh!F F*CJaJ'Hh!Fh.FCJOJQJ^JaJ5h.FhCJOJQJ^JaJ F*CJaJTh.Fhh.FCJOJQJ^JaJcHdhdhdh"F F*CJaJ'Hh"Fh.FCJOJQJ^JaJ`s ttMX$ & F @dEƀJ&.^a$gdX$ & F @dEƀJ&.^a$gds(t)t*t2t3t4t5tZttttХfH-5h.FhCJOJQJ^JaJ F*CJaJ;h.Fh6CJOJQJ^JaJ F*6CJaJTh.Fhh.FCJOJQJ^JaJcHdhdhdh!F F*CJaJ'Hh!Fh.FCJOJQJ^JaJTh.Fhh.FCJOJQJ^JaJcHdhdhdh"F F*CJaJ'Hh"Fh.FCJOJQJ^JaJ5h.FhCJOJQJ^JaJ F*CJaJ tttuAuJuKuLuVuWuXubucuպmBmB.'Hh!Fh.FCJOJQJ^JaJTh.Fhh.FCJOJQJ^JaJcHdhdhdh"F F*CJaJ'Hh"Fh.FCJOJQJ^JaJ5h.FhCJOJQJ^JaJ F*CJaJ;h.Fh6CJOJQJ^JaJ F*6CJaJ5h.FhCJOJQJ^JaJ F*CJaJTh.Fhh.FCJOJQJ^JaJcHdhdhdh!F F*CJaJ tAuuMX$ & F @dEƀJ&.^a$gdX$ & F @dEƀJ&.^a$gdcudueuuuuuuuuպmB.'Hh!Fh.FCJOJQJ^JaJTh.Fhx0h.FCJOJQJ^JaJcHdhdhdh"F F*CJaJ'Hh"Fh.FCJOJQJ^JaJ5h.Fhx0CJOJQJ^JaJ F*CJaJ;h.Fh6CJOJQJ^JaJ F*6CJaJ5h.FhCJOJQJ^JaJ F*CJaJTh.Fhh.FCJOJQJ^JaJcHdhdhdh!F F*CJaJ uuubvvvvvvvպmB.'Hh!Fh.FCJOJQJ^JaJTh.Fh,(3h.FCJOJQJ^JaJcHdhdhdh"F F*CJaJ'Hh"Fh.FCJOJQJ^JaJ5h.Fh,(3CJOJQJ^JaJ F*CJaJ;h.Fhx06CJOJQJ^JaJ F*6CJaJ5h.Fhx0CJOJQJ^JaJ F*CJaJTh.Fhx0h.FCJOJQJ^JaJcHdhdhdh!F F*CJaJ uvwMX$ & F @dEƀJ&.^a$gdX$ & F @dEƀJ&.^a$gdvvvvw\wwwwwwպ~cO$cTh.Fhh.FCJOJQJ^JaJcHdhdhdh"F F*CJaJ'Hh"Fh.FCJOJQJ^JaJ5h.FhCJOJQJ^JaJ F*CJaJ;h.Fh:6CJOJQJ^JaJ F*6CJaJ;h.Fh,(36CJOJQJ^JaJ F*6CJaJ5h.Fh,(3CJOJQJ^JaJ F*CJaJTh.Fh,(3h.FCJOJQJ^JaJcHdhdhdh!F F*CJaJ wwwwwww-x6x7xiN:'Hh"Fh.FCJOJQJ^JaJ5h.Fh3|CJOJQJ^JaJ F*CJaJ;h.Fh:6CJOJQJ^JaJ F*6CJaJ;h.Fh6CJOJQJ^JaJ F*6CJaJ5h.FhCJOJQJ^JaJ F*CJaJTh.Fhh.FCJOJQJ^JaJcHdhdhdh!F F*CJaJ'Hh!Fh.FCJOJQJ^JaJ w-xxxxAyByM>>>>$ h@da$gdJ%X$ & F @dEƀJ& .^a$gdX$ & F @dEƀJ& .^a$gd7x8xExFxGxHxxxxxxxxxxxպպmPm5m5m5h.Fh:CJOJQJ^JaJ F*CJaJ8h.Fh CJH*OJQJ^JaJ F*CJaJ5h.Fh CJOJQJ^JaJ F*CJaJ;h.Fh3|6CJOJQJ^JaJ F*6CJaJ'Hh"Fh.FCJOJQJ^JaJ5h.Fh3|CJOJQJ^JaJ F*CJaJTh.Fh3|h.FCJOJQJ^JaJcHdhdhdh"F F*CJaJxxxxx'y@yAyByCyGyνlL,?h:hhY!CJOJQJ^JaJcHdhdhdhe&?h:hh_FCJOJQJ^JaJcHdhdhdhi&?h:hSh_FCJOJQJ^JaJcHdhdhdhi& h:hCJOJQJ^JaJh:h\cCJOJQJ^J h:h:CJOJQJ^JaJ h:h\cCJOJQJ^JaJ h:h`U6CJOJQJ^JaJ?h:hQhY!CJOJQJ^JaJcHdhdhdhe& ByCyDyEyFyGyHyVyVzXzZzzzz${zK p#EƀF&gdXQdgdG  &gdG $ h@da$gdgoFd&$ h@da$gdJ%GyHyMyQyUyVyWy]y^y`yaybyyӽs\sD.HhFhiGg0J>*CJOJQJ^JaJ,h7-0J>*CJOJQJ^JaJmHnHu'hG hR~80J>*CJOJQJ^JaJ0jhG hR~80J>*CJOJQJU^JaJ h:h+h:hhgcHdhdhdhF+h:h+hgcHdhdhdhF+h:h hgcHdhdhdhF+h:hhgcHdhdhdhF yyzzz.z4zDzJzTzVzZz\zjzlzzz~zzzzzzzzİİİİqqqqqMFjhZ\`hR~80J>*UF*jCJOJQJU^JaJ+hG hR~8>*CJOJQJ^JaJmHsH%hR~8>*CJOJQJ^JaJmHsHhR~8#hG hR~8>*CJOJQJ^JaJ'hG hR~80J>*CJOJQJ^JaJ.HhFhiGg0J>*CJOJQJ^JaJFhG hR~8hiGg0J>*CJOJQJ^JaJcHdhdhdhFzzzzzzz${({,{sP39HhF&hXQhXQCJOJQJaJmH nHsH tHEjHhF&hXQhXQCJOJQJUaJmHnHsH tHu9HhF&hXQhXQCJOJQJaJmH nHsH tHhR~8+hG hR~8>*CJOJQJ^JaJmHsH,h7-0J>*CJOJQJ^JaJmHnHuFjhZ\`hR~80J>*UF*jCJOJQJU^JaJ4hZ\`hR~80J>*F*CJOJQJ^JaJ ,{0{{{{{{{{{{ħo^Z@."HhFhiGg6CJOJQJ"Hh.Fh6CJOJQJ2jHh1Fh6OJQJUmHnHuhXQ!hXQhXQF&*mH sH 1HhF&hXQhXQCJaJmH nHsH tH<HhF&hXQhXQ5CJ$OJQJaJmH nHsH tH9HhF&hXQhXQCJOJQJaJmH nHsH tH9HhF&hXQhXQCJOJQJaJmH nHsH tH<HhF&hXQhXQCJH*OJQJaJmH nHsH tH ${{{{{bM )C$Eƀ.FgdK p#EƀF&gdXQ{{|  ]ݴpfb^ZKGKGhR~8hJhR~8CJaJmHsHh7-h37hiGgHhFhiGg2HhFhX=PhiGg56CJH*OJQJaJ/HhFhX=PhiGg56CJOJQJaJ"HhFhiGg6CJOJQJ%HhFhiGg56CJOJQJ'HhF j&hiGg56OJQJU!HhFhiGg56OJQJ"HhFhiGg6CJOJQJ ( H. Schulz Ed. 10th INTERNATIONAL SCIENTIFIC CONFERENCE ON PRODUCTION ENGINEERING CIM2005 Croatian Association of Production Engineering, Zagreb 2005 B A Picture of environment Path planning World model building Picture analysis and obstacle recognition C D E front back left right through left through right back right back left A very close B close C medium D far E very far MOVEMENT {]gdO"$a$gdhgdtVJ$C$EƀFa$gdiGgˆш҈Ո %&'+ҽ֒}}nnn[E[+h-hR~86CJOJQJ^JaJmHsH%hR~86CJOJQJ^JaJmHsHhJhR~8CJaJmHsH(h'bhR~8CJ OJQJ^JaJ mHsH(h'bhR~8CJOJQJ^JaJmHsH+h'bhR~85CJOJQJ^JaJmHsH(h'bhR~8CJOJQJ^JaJmHsHhR~8(h'bhR~8CJOJQJ^JaJmHsH(hYhR~8CJOJQJ^JaJmHsHˆȈш҈ӈԈՈ &',-239gd%NgdtV$a$gdi$a$gdh+,-239:GHUVWabclmnq}~zs h:hhyo%hR~85CJOJQJ^JaJmHsH+hdjhR~85CJOJQJ^JaJmHsH(h-hR~8CJOJQJ^JaJmHsH%hR~86CJOJQJ^JaJmHsH+h[_BhR~86CJOJQJ^JaJmHsHhR~8+h-hR~86CJOJQJ^JaJmHsH"9:GHVWbcmn}$ h@da$gdgoFd&gdUF(gd%NN 0&P1h:p/ =!"#$% C/F92G$$If!vh5y$#vy$:Vl t65y$$$If!vh5555b55Y5#v#v#vb#v#vY#v:Vl t6555b55Y5$$If!vh5555b55Y5#v#v#vb#v#vY#v:Vl t6555b55Y5G$$If!vh5y$#vy$:Vl t65y$$$If!vh5555?55>5#v#v#v#v?#v#v>#v:Vl t65555?55>5$$If!vh5555?55>5#v#v#v#v?#v#v>#v:Vl t65555?55>52 9G$$If!vh5y$#vy$:Vl t65&$$If!vh5555Q5555#v#v#v#vQ#v#v5#v:Vl t6555555B$$If!vh5555Q5555#v#v#v#vQ#v#v5#v:Vl t6555555B2 92 92 9@@@ NormalCJ_HaJmH sH tH DA@D Default Paragraph FontRi@R  Table Normal4 l4a (k@(No List4 @4 `Footer  !.)@. ` Page Numberj@j B8 Table Grid7:V0PC@"P vBody Text Indent hh`htH4@24 G Header  _$H@BH ga Balloon TextCJOJQJ^JaJ #5Nz} _rXW-,+*(YZ[_`abcdefg| #5Nz}   _rSS/U0UVVVVXX`rOy00 %@0Oy00 4&@0Oy00 %@0 Oy00 &P2#Oy00x$:M@0XY0R-.?N[ghL~  t $D a#c#d#e#f#j#k#n#o#p#q#r#s#v#w#x#y#z#}############!&"&#&$&)q+.,,,,,,,,,,,,/1334444444444444444449<X@Z@\@]@`@b@c@d@e@f@g@h@i@l@m@n@o@p@q@r@s@t@u@v@w@y@{@|@}@~@@@@@@yDzDEGHfM7O)P{S|SSSSSSSSSSTTTT3T6TJTNTfTgTiTkTmToTqTsTuT|TTTTTTTTTTTTTTTTTTTTTTTTUUUU/U0U1U:UDUIUTUcUzU{U}UUUUUUUUUUUUUUUUUUUUUU VVVV(V5V9VCVNV[VdVpV{VVVVVVVVVVVVVWW%W6WAWHWTW[W\W^W`WbWdWfWhWjWWWWWWWWWWWWWWXXX)X7X?XLXXXdXnXxXXXXXXXXXXXYZi_`aacbcpcfgghh`i jjAkklm-nnnnAoBoCoDoEoFoGoHoVoooo#p$p%p=ppppppqEqFqGqHqJqKqMqNqOqPqgqhqiqjqkqlqmqsq|q}q~qqqqqqqqqqqqqqqqqqqqqqqqqrr rrrr(r2r=rErRrSr\r]r`r00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 0 0 0 0 0 0 0 0 00000000000000000000000000000000000000000000000000000000000000000000000000000 0 0 0 00 00 00 00 0 0 0000000 0000000 000 0000000 0000 0 0000 0 00 0 0 0 0 0 0 0 0 0 00000000 000000 00000000 000000 000000 0000 0000 0 00 0 0 0 0 0 0 00 00 0 0000000 0 000000 000000 00000 000000 000000 0 00000000000000 0 0 0 0 0 0 0 0 0 0 00000000000@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0 0@# 00000000000000000000000000000000000000000000000000000000-.?N[ghL~  t $D a#c#d#e#f#j#k#n#o#p#q#r#s#v#w#x#y#z#}############!&"&#&$&)q+.,,,,,,,,,,,,/1334444444444444444449<X@Z@\@]@`@b@c@d@e@f@g@h@i@l@m@n@o@p@q@r@s@t@u@v@w@y@{@|@}@~@@@@@@yDzDEGHfM7O)P{S|SSSSSSSSSSTTTT3T6TJTNTfTgTiTkTmToTqTsTuT|TTTTTTTTTTTTTTTTTTTTTTTTUUUU/U0U1U:UDUIUTUcUzU{U}UUUUUUUUUUUUUUUUUUUUUU VVVV(V5V9VCVNV[VdVpV{VVVVVVVVVVVVVWW%W6WAWHWTW[W\W^W`WbWdWfWhWjWWWWWWWWWWWWWWXXX)X7X?XLXXXdXnXxXXXXXXXXXXXYZi_`aacbcpcfgghh`i jjAkklm-nnnnAoBoCoDoEoFoGoHoVoo=ppppppqEqHqqrr`r00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 0 0 0 0 0 0 0 0 000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 0 0 0 0 0 0 0 0 0 0 000000000000My00My00zMy00@0My00@+3My00My00My00@00N O900TO900TO900T0:  ry9A+x g ^#7$%&'_*+h--.#0E2356/79;=W>??@YBDFVH}IYJJ*MNOQnRS"VWoY)Z$[[-]]L^^0__9``aab2cd]fh.jkmprrstcuuvw7xxGyyz,{{+>ABCDFGHIJKLMNOQRSTUVWXY[\]^`abcdehlmnoprstuvwxyz|~%-66666>dJJ{]]f^g^^__z_{_(```[a\aLbbbr`stuwBy${{9?EPZ_fgijkq{}@ !!@  ~@d 0}(  2 * 6*>"` 2 < B2KENHQԔ `T2K.`T2K.`T2K.`T"`2 = BhJCXPENGȷHdJXPQԔ `ThJ(`ThJ(XPhJ(XP"`2 > ,B NC KENG2HJ KQԔ `Tl& N*`Tl& N* K& N* K"`2 ? PB.VENGmH+IQԔ Le.V/$Le.V/$`T.V/$`T"` 2 @ śBhJCMENGoHdJMQԔ `T]4 hJ5&`T]4 hJ5&M4 hJ5&M"`2 A ޒBhJCJENGkHdJJQԔ r`T2&hJ #r`T2&hJ #J&hJ #J"`2 ( 6(>"` 2 + 6+>"`  2 - 6->"` t _ s * PF0*PF0*_"`   t ` s * PF0*PF0*`"`  t a s * PF0*PF0*a"`  t b s *PF0*PF0*b"` t c s *PF0*PF0*c"` t d s *PF0*PF0*d"` t e s *PF0*PF0*e"` t f s *PF0*PF0*f"` t g s *PF0*PF0*g"`  @ v (3 \ P2 D  g "`* J'3P2 N  g "`'G,#P2 M  g "`>"F)P2 L  g "`[$+'fB HB s *Dg 0:3ZB Q S Dg 3ZB R S Dg {3ZB T S Dg  % X s *PF0*PF0*X"`=%7&  Y s *PF0*PF0*Y"`%&  Z s * PF0*PF0*Z"`"#%$&   [ s * PF0*PF0*["`'%(&   W s *PF0*PF0*W"`x%T& ZB G S D1v %'%V2 K # "`p%$&(ZB S S Dg C S'%NB r S DNB s S DNB t S DNB u S DNB v S DNB w S DNB x S DNB y S Dj@ #O" }2 , 6,>"`#O"  | 0PFжPF|"`@#E B S  ?o`<=>?@AJ2(  NB  S DoNB  S D1NB  S D1a#f#g#h#k#l#s#t#z#{#}#X@Z@]@^@`@i@j@w@y@@uT|TTTTTTT_r(Dt= tAWVt< t*8 ft-t@S!t> 6t}9]t+Wgt?&{at_4&tt\O/#tgQHtag tb !td!.'tcC\tfo tep.$>t`!(atrS,tsett{Otut,twJCtv%tybp3txz-t:t"ut"u\#\u#<|<|_,` l` ` ` ,a laaWO|,d <| <  QTQQ\ {!{%2EE^^lyiijjjjnnn8q8q`r      !"!++4KKhrr{jjjjjjnnn>q>q`r   !"8*urn:schemas-microsoft-com:office:smarttagsCity=!*urn:schemas-microsoft-com:office:smarttags PlaceName="*urn:schemas-microsoft-com:office:smarttags PlaceType9#*urn:schemas-microsoft-com:office:smarttagsplaceB *urn:schemas-microsoft-com:office:smarttagscountry-region #"!###"!###"!###"!### # ###mntu~d x ##$&%&(&)&33334444555555Y@@@@@@@@DDDD7OOOOO{SSWWWWWWBXEXXXXXXXXYYY Y YYYYY1Y2Y:Y;Y?YAYDYEYMYNYXYYYcYdYkYlYtYuYYYYYYYYYYYYYYYYYYY Z ZZZZZ1Z2Z6Z7Z9Z:Z=Z>ZFZGZLZMZOZPZ`ZaZgZhZZZZZZZZZZZ[[ [ [[[\[][b[c[o[p[r[s[z[|[[[[[[[[[[[[[[[[[[[[[[[[[[[\\\\\\\\a\b\l\m\u\v\\\\\\\\\\\\\\\\\]]] ]]]]]]]!]"](])]0]1]4]5]9]:]?]@]F]G]O]P]R]S]V]W][]]]a]b]e]f]]]]]]]]]]]]]]]]]1^2^6^7^<^>^A^B^J^K^M^N^Q^R^Z^[^]^^^a^b^f^g^l^m^r^s^^^^^^^^^^^^^ _ _____3_4_:_;_A_B_I_J_M_N_V_W_Y_Z_`` ` `````-`.`2`3`9`:`D`F`G`H`|`}`````````````````````````````aaaaaaBaCaKaLaNaOaSaTaVaWa^a`aaaaaaaaaaaaaaaaaaaaaabb bbb_b`bgbhbqbrbbbbbbbbbbbbbbbbbbbbb c ccccccc8c9c^h^`OJQJo(hHh^`OJQJ^Jo(hHohpp^p`OJQJo(hHh@ @ ^@ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohPP^P`OJQJo(hH^`o(. ^`hH. pLp^p`LhH. @ @ ^@ `hH. ^`hH. L^`LhH. ^`hH. ^`hH. PLP^P`LhH.H`'T                  Z^--|t1%JYOS_`& d ~  .: R r : ; 0I W4 ;s 10CuR,J% @-r8zIoY|d4GY3I=c_*7yAdi!<9hbl] 3&!Y!"C"O"O8#O#`#v#/$%l%)&X'UF(L(t(z)5*1*8*9+w+i,.+,B,GL,7-@-.X&.v/I5/R?/T/NL0x0OD1q12F2J3,(3O3m34H4- 5M5ao5>6`U6y6}6w7R7?8I'8B8I8R~8 :+<=,=t=m9@@@MATQA[_B`BCECwCGDjDkDEzE FCF?WF_FGv$GRGcGdH:pI;J_JK6KEKLbK;LwN%NKN(EPKjP sPvQb'R[_R5T_>j__` `Z\`Ck`4a8c\cscwcdddcefQfefiGgugayin$lm$n^o}o7pdrK s bs1ZuvrlvxLyhz@z1kz!{3||H|W|to|n}G~e~;2)'-DGGUr wd1e!mqSBU/ bu|:XV!h3N7FhJ 37Np|gaNGKh@`yXQd i7k>U^axP92.FLbv%%4am3uxv dhPXt fD%*vGL[n,SJ*S#<k-9 7Qw1$ GZ?&.v1T }j"S'b vwdjSy0Q$lB!Nj[fjojrj|*x dSSSSSTT6TNTfTgTuTTTTTTUUU/U0U1U:UDUIUTUcUzU{UUUU5VpVVVVVVVVWW%W6WHW[W\WjWWWW7XxXXX`r""""@SSSS@op_rpp p@pzp@UnknownMladenTehni ki strani jeziciMamaGz Times New Roman5Symbol3& z Arial5& z!Tahoma?5 z Courier New;Wingdings#1hh4F2FGy^8^8!24doo 3QH)?1%/Kognitivna rabotika i planiranje putanje robotaMladenMladen  Oh+'0  ,8 T ` l x0Kognitivna rabotika i planiranje putanje robotaognMladenvladladNormalvMladenv71dMicrosoft Word 10.0@4@| M4@('#0@iUM4^՜.+,D՜.+,d  hp   Crnekovica8oA 0Kognitivna rabotika i planiranje putanje robota TitlelH(DX`_AdHocReviewCycleID_EmailSubject _AuthorEmail_AuthorEmailDisplayName_PreviousAdHocReviewCycleID_ReviewingToolsShownOnce\mTekst-konačna verzijaegrgur.tokic@fsb.hri Grgur Tokic*yrgu  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$%&'()*+,-./012345689:;<=>@ABCDEFKRoot Entry FpaaM4MData 1TableRWordDocumentPSummaryInformation(7DocumentSummaryInformation8?CompObjj  FMicrosoft Word Document MSWordDocWord.Document.89q