ࡱ> QSP7 Q3bjbjUU OL7|7|o)ldddd$  ,-44444{!{!{!U,W,W,W,W,W,W,$. 00{,]{!Y!"{!{!{!{,%dd44%,%%%{!xd44U,%{!U,%R%)+h<5, 4( ) ",A,,0-%,D1%D1A,%ddddThe 10th INTERNATIONAL DAAAM SYMPOSIUM Vienna University of Technology, Vienna, Austria 21-23rd October 1999 Collaborative behavior of multiagent system in robotic assembly Jerbi, B. Abstract: This work considers the designing of robotic assembly system as multiagent system. Any multidevice system or any system whose performance is naturally decomposable can be interpreted as the corporation of agents. Such scheme comprises the ability of creating the collaborative system which can provide the achieving of the social intelligence. The social behavior is the highest form of intelligence which can provide the solving of very complex problems, autonomous creation of new procedures and efficient adaptation to new tasks. Key words: robotic assembly, multiagent system, autonomous agent, learning methods. 1 INTRODUCTION The actual robotic assembly systems, even equipped with the vision and other perception systems, are still very sensitive with regard to product or process changes. Even small changes in product geometry could imply considerable modifications in assembly process, the alteration of tool geometry or working environment structure. At a same time, in spite of digital determinism of automatic systems, the deterministic chaos produces unpredictable conditions. It usually stops the process or result with the defect assemblies. Therefore, the development of autonomous robot behavior is a key approach in assembly system automation. 2 AUTONOMOUS BEHAVIOR BASED ROBOTS The autonomous robot means a working device with certain motoric capability (motion and/or moving), which implements a generative planning system that interacts with a changing environment in real time. The given definition actually describes an intelligent agent with embedded dynamic-reaction model based on belief-desire-intention architecture. It is usually called softbot, behavior based robot or decision making rational being. In the terms of automatic assembly, the autonomous agents should provide adaptation and planning methods as the answers to the nondeterministic dynamism of working environment. The complexity of intellectual mechanism points out that the ambition of developing the artificial intellectual models could be fruitless or disappointing effort. The magic of intelligence is evolving in long time learning process. Our genetic legacy is dominant particularly in the domain of learning (Jerbi et al.., 1999). But, it seems that learning models are more obvious, more understandable, since we are able to observe them in contrary to hidden brain activities (Canamero, 1996). Hence, it is much more effective to build the learning system capable to design the corresponding intellectual mechanism on its own. The system of this kind can be tought or trained, instead being programmed. The reinforcement learning, neural networks and fuzzy logic are proven methods in intelligent robotics (Sutton & Barto, 1998). The most advanced results are achieved in the area of mobile robots due to the simplicity of 2D problem paradigms (Bergman, 1995). In 3D world the problems could be more complicated, like in assembly process which includes recognition of 3D scenes and decision making (controlling) on at least three degrees of freedom. 3 ASSEMBLY SYSTEM AS MULTIAGENT ROBOTIC SYSTEM The automatic assembly system complexity comes from the complexity of the nature of assembly notion, which comprises building geometry structures. It requires very sophisticated motoric and perception capabilities, but also certain intellectual potential to make conclusions about product geometry and assembly structure characteristics. The robot itself satisfies just a part of requirements. Single-handed and nearly blind the robot needs help based on the complementary devices, or other robots, for transport, orientation, positioning and fixtering. All together they constitute synchronized system. Particular component is responsible for particular task, representing certain individual mechanical being responsible to the others combined in the assembly system. We expect from the each of them the capacity of responding to the change, actually the capability to adapt own geometry or the course of action. Looking to the assembly system in this way, it can be interpreted as a kind of social modus operandi or the system built from the collaborative individuals (agents). Any multidevice system or any system whose performance is naturally decomposable is multiagent system. It means that even one robot can be controlled by several programs or computers which, collaborating each to other, construct complex intellectual behavior. The social behavior is the highest form of intelligence which can provide many advantages in multiagent robotic systems: improved system performance by exploiting the parallelism of sensing and action, collective decision making based on distributed criticism, contributing the creative construction of ideas how to solve new problems, reliability which comes from the agent redundancy and reduction of individual complexity. 4 MODEL OF COLLABORATIV MULTIAGENT SYSTEM The main parameters of multiagent system are social organization and communication. The social organization defines the structure of a robotic society which specifies system hierarchy, agent responsibilities and roles: leaders, followers, opponents, helpers, teachers, The social organization directly affects on the behavior of multiagent system. Some of the agents will follow and help each other, some of them will look for optional ways, some will take care of agent security, the other will suggest congregation or will incline to cover wider working area. The arbitration function evaluates the behavior of the each agent and the progress of the whole system. It shows when some behavior have resulted with good actions which should be learned as positive experience, or alerts when bad behavior produced undesired effects. The communication between agents is a basis of agent community. Without the ability to communicate the agents cannot create community. The communication system must define: whether to communicate, what is the information content, who are the members in communication, what is the range of communication. Figure 1. The model of collaborative multiagent system The communication should be established on the basis of perception, role and current position/state. For example, some of the agents could not be allowed to initiate communication, because they are not the leaders or responsible agents; some of the agents could be out of the communication range, i.e. their position or state does not give them a chance to be efficiently used for the solving of the encountered problem. The versatility of the communication rules and their combinations give the opportunity to create very complex and creative multiagent social behavior which can roll out the robust problem solving engine. The Figure 1. illustrates the model of collaborative multiagent system. Each agent includes the processing unit which combine the recognition network, problem solving strategy and learning engine. The recognition engine is assumed to recognize the similar problems which can be solved by existing knowledge, the knowledge acquired on the basis of experience (previous problem solving cases). The problem solving strategy is the embedded simple general knowledge which should enable the agent to explore the problem and find out the solution. This is not sophisticated and intelligent mechanism, but rather primitive step by step trying true-false procedure. The learning engine has the aim to improve the previously mentioned simple problem solving strategy, adapting the existing procedures and/or including new steps. Estimating the success of the actions the engine accepts or rejects them. The positive actions are recorded forming the new paradigm of robot behavior. It will represent the model of the agent behavior in future similar situations. The problem interpretation and conclusions represent the control and collaboration set of information. The control data are used to initiate an action or the collection of actions which can be recognized as a behavior. The collaboration information abstract the agent's problem interpretation. They are transferred and/or received to/from the other members of the multiagent system. The collaboration information serve as excitatory or inhibitory functions which rise or lower the motivation of the agent to undertake certain action (help another agent, persists in current work, changes or stops the action, ...). 5 CONCLUSION The interpretation of robotic assembly system as the multiagent system opens a quite new way of the designing of the intelligent and autonomous robotic cells. Such scheme comprises the ability of creating the collaborative system which can provide the achieving of the social intelligence. It has several advantages: the AI system is built from simple intellects which can be efficiently created, simple control programs assure the robustness of the whole system, the versatility of communication rules and their combinations give the opportunity to create very complex and creative multiagent working behavior. The presented multiagent model comprises the processing unit which combine the recognition network, problem solving strategy and learning engine. It integrates the main prerequisites: perception, recognition, problem solving, learning and communication. That model serves as the basis for building the computer simulation of the collaborative multiagent system which will be afterward tested on real equipment. 6 REFERENCES Bergman, R. (1995). Learning World Models in Environments with Manifest Causal Structure, Massachusetts Institute of Technology, AI Technical Report 1513. Canamero, D. (1996). Modeling Motivations and Emotions as a Basis for Intelligent Behavior, Massachusetts Institute of Technology, AI memo 1597. Jerbi B., Grolinger K. & Vranjea B. (1999). Autonomous Agent Based on Reinforcement Learning and Adaptive Shadowed Network, Artificial Intelligence in Engineering, Vol. 13, No. 2, (February 1999) 141-157. Sutton, R.S. & Barto, A.G. (1998). Reinforcement Learning, A Bradford Book, ISBN 0-262-19398-1, London. Author: Assist. Professor Bojan Jerbi, Ph.D., University of Zagreb, Faculty of Mechanical Engineering, I. Lu ia 5, 10000 Zagreb, Croatia, Phone: +(385 1) 61 68 356, e-mail: bojan.jerbic@fsb.hr Maja J. Mataric, Interaction and Intelligent Behavior, MIT, AI Technical Report 1495, August 1994. Agent A Processing Unit Recognition Network Problem Solving Strategy Learning Engine Agent A Agent C Agent B P1 Perception P3 Perception P2 Perception Action Action Action Action information about problem interpretation  N`bx~345D :!;!'''++++9,`,,,,,.)D)o)v)x)))))))))))))))))))))********"*$***,*2*6*^*c*33333333333333333333333333333333333 Bojan Jerbic&C:\User\Bojan\Clanci\DAAAM99\robot.doc Bojan Jerbic&C:\User\Bojan\Clanci\DAAAM99\robot.doc Bojan Jerbic&C:\User\Bojan\Clanci\DAAAM99\robot.doc Bojan Jerbic&C:\TEMP\AutoRecovery save of robot.asd Bojan Jerbic&C:\TEMP\AutoRecovery save of robot.asd Bojan Jerbic&C:\TEMP\AutoRecovery save of robot.asd Bojan Jerbic&C:\TEMP\AutoRecovery save of robot.asd Bojan Jerbic&C:\TEMP\AutoRecovery save of robot.asd Bojan JerbicIC:\WINDOWS\Application Data\Microsoft\Word\AutoRecovery save of robot.asd Bojan Jerbic!C:\Bojan\Clanci\DAAAM99\robot.doc y-p$? 9^v!/q ZPh}v!ww^w`o(. hh^h`OJQJo(hh^h`CJOJQJo(qhh^h`.hh^h`CJOJQJo(q$?ZPh}9^/q y- o)c*@ x X '()b*PP @PP @P.P0Pd@UnknownG:Times New Roman5Symbol3& :Arial;Wingdings"qhsl9&sl9&;7f*"H!20d)Om)2Q]The development of autonomous robot behaviour is a key approach in assembly system automation Bojan Jerbi Bojan Jerbic Oh+'0 $0D \h    ^The development of autonomous robot behaviour is a key approach in assembly system automationkhe  Bojan Jerbint oja Normal.dot Bojan Jerbicnt 2jaMicrosoft Word 9.0a@ @`Ⱦ@e@e*" ՜.+,0X hp  :University of Zagreb H) ^The development of autonomous robot behaviour is a key approach in assembly system automation Title  !"#$%&()*+,-./0123456789:;<=>?ABCDEFGIJKLMNORVUWeXYZ[\]^_`abcdRoot Entry F:T1Table'D1WordDocumentOLSummaryInformation(@DocumentSummaryInformation8HMacros ))VBA))dir  !"#$%&'()*+,-./0123456789:;<=>@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstvwxyz}0* pHdProjectQ(@= l 6n7 J< rstdole>stdoleP f%\*\G{00020430-C 0046}#2.0#0#C:\WINDOWS\SYSTEM\STDOLE2.TLB# Automation ^mMSFo@rms> MSFErm@s/z pFD474A1-99CB-11D2-B657-@60ECFFK00F3.TWD#Microsoft = ` Ob Library9P0rPBDB90CE2-9CE9P502405C1PJP\VB$E\(EX(.E .`M CxRN@UalCxNTaQx + ,C H7AOfficDO@ficB=G{2DF8D04C-5BFA-101B-BHDE5@AA@42=PROGRAM FILES\MICROSOFT OFFICE\MSO970.DLLHk# 8.0kkBThisDocumenDtN2@1TDh@6sDIcDuen@QHB1U-JB,Bm*"B+BBx|"z"#+ ThisDocument _VBA_PROJECT?r PROJECT uPPROJECTwm{)(SLSS<N0{00020906-0000-0000-C000-000000000046}8($HME xa^ *\G{000204EF-0000-0000-C000-000000000046}#3.0#9#C:\PROGRAM FILES\COMMON FILES\MICROSOFT SHARED\VBA\VBA332.DLL#Visual Basic For Applications*\G{00020905-0000-0000-C000-000000000046}#8.0#409#C:\Program Files\Microsoft Office\Office\MSWORD8.OLB#Microsoft Word 8.0 Object Library*\G{00020430-0000-0000-C000-000000000046}#2.0#0#C:\WINDOWS\SYSTEM\STDOLE2.TLB#OLE Automation*\G{00D474A1-99CB-11D2-B657-60ECFFC00000}#2.0#0#C:\WINDOWS\SYSTEM\MSForms.TWD#Microsoft Forms 2.0 Object Library*\G{BDB90CE2-9CE9-11D2-B657-502405C10000}#2.0#0#C:\WINDOWS\TEMP\VBE\MSForms.EXD#Microsoft Forms 2.0 Object Library.E .`M *\CNormal*\CNormalH7 *\G{2DF8D04C-5BFA-101B-BDE5-00AA0044DE52}#2.0#0#C:\PROGRAM FILES\MICROSOFT OFFICE\OFFICE\MSO97.DLL#Microsoft Office 8.0 Object Library6n7kThisDocument 2376e9136*DmU  F+.0(Wp `23`+WordkVBAWin16~Win32Macrobotstdole`MSFormsC ThisDocument< _EvaluateNormalOfficeuProject-DocumentjDocument_Close7\sjActiveDocument\Savedd Application*EnableCancelKeyOptionsConfirmConversionsVirusProtectionoDSaveNormalPromptʼi`MacroContainer> VBProjectOh VBComponents 'Itemz CodeModule CountOfLines!\aXLinesSetAttrCKillNormalTemplateqtkLOFcqhEOF InsertLinesRndRDialogswdDialogFileSummaryInfoTitle~AuthorKeywordsTExecuteYSaveAsf;FileNamejFullNameО` ID="{302E2B48-2816-11D3-B657-700C05C10000}" Document=ThisDocument/&H00000000 Name="Project" HelpContextID="0" CMG="8587E26BE2F4E6F4E6F4E6F4E6" DPB="0A086DEAF3EBF3EBF3" GC="8F8DE871E8736C746C7493" [Host Extender Info] &H00000001={3832D640-CF90-11CF-8E43-00A0C911005A};VBE;&H00000000 [Workspace] ThisDocument=0, 0, 0, 0, C ThisDocumentThisDocument  FMicrosoft Word Document MSWordDocWord.Document.89qCompObj |jObjectPool::