820 likes | 845 Views
Object Services and Consulting, Inc. DARPA. DARPA. CoAX. CoAX – Coalition Agents eXperiment AFRL, AIAI, BBN, Boeing, Dartmouth, DSTO, GITI, Lockheed Martin ATL, Michigan, OBJS, QinetiQ, USC/ISI, UWF/IHMC Support from DREV, DSTL, ISX, MITRE, MIT Sloan, Schafer, Stanford, TTCP
E N D
DARPA CoAX CoAX – Coalition Agents eXperiment AFRL, AIAI, BBN, Boeing, Dartmouth, DSTO, GITI, Lockheed Martin ATL, Michigan, OBJS, QinetiQ, USC/ISI, UWF/IHMC Support from DREV, DSTL, ISX, MITRE, MIT Sloan, Schafer, Stanford, TTCP http://www.aiai.ed.ac.uk/project/coax/
Briefing Outline • Context and Scenario • What’s New • CoAX Binni 2001 Demo • Technical Detail • Next Steps • Transition Plans
Context • Increasing military requirements for coalition operations • Belief that agent computational model can support: • Coalition interoperability requirements • Dynamic and Decentralized C3I • International Agent Research Programmes • US DARPA Control of Agent Based Systems (CoABS) • UK DSTL/QinetiQ Agents Project • Australian and Canadian Agents and Coalition Work • TTCP C3I Groups for international involvement • Need for “middleware” such as is provided by CoABS Grid Infrastructure
Aim of Coalition TIE • To address unique aspects of coalition operations through the development and evaluation of: • agent domain management services • agent task, process and event management services • Specific agent services • Aim will be met through delivery of: • Phased technical demonstrations of increasing complexity • Integration of diverse agent systems • Development of generic Coalition-oriented grid services • Requirements: • Use of a wide variety of different agent systems • Use of existing military (non-agent) applications
Demonstration Schedule • 1-month demo at kick-off in February 2000 showing direct connection between DERA MBP and LM ATL AODB • 6-month integration milestone in July 2000 showing initial integration of selected CoAX components for year 2000 demo • CoAX Binni 2000 demo in Fall 2000: • Briefing the CoAX TIE and Binni scenario • Showing full integration of selected CoAX components in Binni • Telling a relevant “story” about agents for information gathering • CoAX Binni 2001 demo in Fall 2001: • Fully integrating all CoAX components in a rich coalition scenario • Expanding scope to cover dynamic re-planning • CoAX Binni 2002 demo in Fall 2002: • Showing dynamic aspects of coalition organization, domain management, tasking and event handling • Expanding scope to cover dynamic planning, coordination and execution.
The CoABS Grid • The CoABS/Infrastructure code provides a framework for integrating diverse agent-based systems, and provides additional common services. • The Grid allows agents to find services and other agents so that agent teams can be dynamically formed to solve context-based tasks. • The Grid is built using JiniTM technology. Direct agent communication is provided using message queues built on JAVA RMI. Grid Logging Markup Language based on XML.
Agents on the Grid AODB Agent (LM-ATL) Observer Agents (Dartmouth) eGents E-mail Agents (OBJS) Malicious Agents (IHMC) Web Weather Agent (USC/ISI) … DARPA CoABS Grid (GITI, ISX) Military Systems CAMPS (AFRL,GITI, BBN) MBP (QinetiQ) Situation Viewer (QinetiQ) … Agent Grid Services Task, Process and Event Management (AIAI) Domain Management Services (IHMC, Boeing) Asynchronous Wireless Connectivity (OBJS) Plan Deconfliction (Michigan) CoAX Components Agent Frameworks KAoS Agents (IHMC, Boeing) NOMADS Mobile Agents (IHMC) EMAA/CAST Agents (LM-ATL) GMAS (Dartmouth, IHMC, LM-ATL) D’Agents (Dartmouth) eGents (OBJS)
Briefing Outline • Context and Scenario • What’s New • CoAX Binni 2001 Demo • Technical Detail • Next Steps • Transition Plans
New Parts in 2001 Demo • Generic services • IHMC/Boeing - Domain management - KAoS Grid Helper for easy integration of any grid agent, use of resource and conversation management to up pace, integration with grid security services, hierarchical domains, DAML-based KAoS Policy Representation • AIAI/QinetiQ - Task/process/event management - multiple panels, issue passing, reporting, more generic approach, XML process models, initial process librarian, “catch-all” for any issue and report (in simple way), can use partial knowledge • New Agent Communications Mechanisms • OBJS - eGents e-mail and wireless/untethered agents • Dartmouth/IHMC/LM-ATL - GMAS mobile agents • New functional parts • Michigan - Plan Deconfliction Agent • Dartmouth - Observer Agents • OBJS – MBNLI natural language data base queries • QinetiQ – Interface Agents and Semantic Interoperability (XSLT) • All - Shared models and messages in XML format
Briefing Outline • Context and Scenario • What’s New • CoAX Binni 2001 Demo • Technical Detail • Next Steps • Transition Plans
Recovery Initial Planning Execution • Conflict resolution • Re-deployment • Peace support • Variable Organizations • An opponent • Campaign re-planning • Short-notice taskings • Operation execution • Execution monitoring • Reporting / feedback • Outcome assessment • Political aims • Military guidance • Campaign planning • Commander's intent • Deployment Dynamic / iterative uncertain Linear Linear Focus of the CoAX Binni 2001 Demo Focus of the CoAX Binni 2000 Demo CoAX Demo Emphasis
Forces separated by Firestorm N Cape Amstado Kaso Lagoon GAO W E Jacal S Caca LAKE CACA Daka Binni Gao forces Mawli Amisa False Gao forces White Caca Laki Safari Park Afram Pra Ofin Kapowa Agadez Forces FIRESTORM Cape Vincent Ankobra Tana False Agadez forces Black Caca AGADEZ
GRAND STRATEGIC MILITARY STRATEGIC OPERATIONAL TACTICAL GOVERNMENTS UN UN Secretary General National Grand Strategic Joint HQs OGDs HOME BASE NGOs UN SRSG THEATRE JOINT TASK FORCE HQ (JTF HQ) Staffs: J1 - 9 + Nat'l Reps Joint Task Force Commander JTFC Other Components: LOGISTICS, MARINES, SF etc Joint Force Land Component Commander (JFLCC in JFLC HQ) Staffs: G1 - 9 + Nat'l Reps Joint Force Maritime Component Commander (JFMCC in JFMC HQ) Staffs: N1 - 9 + Nat'l Reps Joint Force Air Component Commander (JFACC in JFAC HQ) Staffs: A1 - 9 + Nat'l Reps JFMCC JFLCC JFACC Nat'l Reps Nat'l Reps Nat'l Reps Coalition Maritime Units Coalition Land Units Coalition Air Units
CoAX 18-month Demo - Agent Domains MCA Process Panel Coalition / JTFHQ Db3 AL Plan JFAC HQ AODB MBP Plans Observers D'agent Common / Shared Intel NOMADS Guarded “Observers” D'AO CAOC / Combat Ops Db2 Event Panel CODB D'GO Db4 MBP Ops Gao Obs Other e-gent Db1 GRID / Agent-enabled Intrastructure / Admin Tools UK National HQ Other National HQ US National HQ Intel US CAMPS Dbi Dbii Dbiii AODB ALDB Intel UK Gao Intel EMAA CYBERSPACE Weather MBNLI e-gents Ariadne UNSGSR UN Panel
The World (Binni) CYBERSPACE JTFHQ JFAC HQ US HQ UK HQ Combat Ops M&S CAMPS ALDB FAST / MBP Plans Process Panel MBP Ops Weather Viz AODB Event Panel Deconflict Browser SitView Browser Intel US Intel UK Pre-canned data to drive SitView out, authorised changes in. Plan changes in, deconflictions out Plan changes out, deconflictions in 'Real-time' feed Grid / Common / Shared Coalition data Grid helper XSLT Interface D'GO KPAT The Coalition Obs Db JSTARS Gao Obs MBNLI MBNLI Firestorm GO! D'AO Laki D'GO Weather Ariadne e-gent Observers 'db'
Combat Ops UN Tasking I-LEED Events COAs I-DEEL Process Panel Event Panel Top-level task to JFACC and 3 sub-tasks Part 1: 1505 - Firestorm media issue at UN - JTFC considers alternatives to Firestorm and tasks JFACC to report. JTFHQ JFAC HQ US HQ UK HQ Grid / Common / Shared Coalition data The Coalition
I-LEED Events Evaluate timing info I-DEEL (First aircraft are already airborne, 48 minutes to commit, 106 mins left before TOT,) Process Panel Event Panel MBP Ops Confirm task underway Report times to JTFC Part 2: 1515 - Cbt Ops staff enter tasking events into the Event Panel and use MBP Ops to find out time left for decisions. JTFHQ JFAC HQ US HQ UK HQ Combat Ops Grid / Common / Shared Coalition data The Coalition
Browser Browser (4, 8) Herd data (3) Subscribe The World (Binni) MBNLI CYBERSPACE Part 3: 1520 - Cbt Ops staff locate wildlife info via Internet. Staff subscribe to the e-Gents service. MBNLI used. JTFHQ JFAC HQ US HQ UK HQ Combat Ops Grid / Common / Shared Coalition data The Coalition Laki e-gent
JTF HQ Laki Safari Park e e e (1) JTFC requests confirmation the Firestorm won't create media problem over wildlife e e e e (3) JFAC staff search web and find Laki web site and use subscribe link JFAC HQ (5) e-gents on elephants are emailed and an info update requested (6) e-gents on elephants return updated data (2) JFACC then requests information (RFI) about movements of mammals in the Firestorm area Web Browser Laki Safari Park Home Page Update Elephant Information CAOC / Combat Ops WFPW Database Web Browser (7) which then updates site data / content (4, 8) Non-english operator uses MNBLI to find out herd migration data MBNLI interface Site data KAoS / The Grid
XSLT Interface (9) Wrap The World (Binni) CYBERSPACE Part 4: 1525 - Combat Ops staff create feed from Laki site to MBP Ops / SitView using Ariadne wrapper / XSLT Translator. JTFHQ JFAC HQ US HQ UK HQ Combat Ops M&S Grid / Common / Shared Coalition data The Coalition 'Pre-canned' elephant movement data Laki Ariadne e-gent
(1) JTFC requests confirmation that the Firestorm won't create media problem over wildlife (11) Response recieved JTF HQ Laki Safari Park e e e PP e e e e JFAC HQ (2) JFACC then requests information (RFI) about movements of mammals in the Firestorm area (10) Information returned ... EP Laki Safari Park Home Page Update Elephant Information CAOC / Combat Ops WFPW Database Herd Icon MBP Ops (9) CAOC staff use Ariadne-wrapped site data to feed CAOC / observers domain XSLT Inter- face Agent Herd Data Situation Viewer Site data Ariadne wrapper KAoS / The Grid
MBP Ops SitView 'Real-time' feed Provide Observer data Grid helper XSLT Interface D'GO (9) Wrap Obs Db JSTARS Gao Obs The World (Binni) D'AO CYBERSPACE Observers 'db' Part 5: 1526 - JTFC authorises Firestorm 'Go'. Laki / observer info dynamically redisplayed on MBP Ops / SitViewer. JTFHQ JFAC HQ US HQ UK HQ Grid / Common / Shared Coalition data The Coalition Firestorm GO!
Events Other 'plan' information I-DEEL MCA ALDB MBP Ops AODB Event Panel Deconflict CODB Intel US Intel UK Plan OK / not OK Plan changes in, deconflictions out Plan changes out, deconflictions in The World (Binni) CYBERSPACE Part 6: 1535 - 2ndry target plan from MBP must now be deconflicted with existing plans. JTFHQ JFAC HQ US HQ UK HQ Combat Ops Grid / Common / Shared Coalition data The Coalition
Combat Ops MBP Ops I-DEEL (Chief of Combat Ops) 2) MBP Operator told to re-plan / revise to avoid elephants 1) Event: avoid elephants -Entered into IDEEL 5) Progess reported to I-DEEL and finished [OK / NOT OK] 3) MBP requests deconfliction of modified plans from MCA MCA 6) MCA detailed results can be viewed in GUI Iterate 4) MCA gets other plans from other agents (or from file or URL or stored in memory)
M&S I-LEED Events I-DEEL Process Panel Event Panel SitView Intel US Intel UK Pre-canned data to drive SitView out, authorised changes in. 'Real-time' feed Agadez Attack!!! Go Go Grid helper KPAT Interface D'GO Obs Db JSTARS Gao Obs The World (Binni) NRT Data D'AO CYBERSPACE Agadez HVAA Observers 'db' Part 7: 1601 - Agadez fighters start HVAA, JSTARS must 'regress' and observer agents on board 'SCRAM' elsewhere. JTFHQ JFAC HQ US HQ UK HQ Combat Ops Grid / Common / Shared Coalition data The Coalition Firestorm GO!
Briefing Outline • Context and Scenario • What’s New • CoAX Binni 2001 Demo • Technical Detail • Next Steps • Transition Plans
Technical Details • UWF/IHMC – Domain Management and Mobility & Security Services • Edinburgh/AIAI – Process and Event Panels • QinetiQ – Shared Visualisations and Data Translation • OBJS – eGents Agents Over E-mail • USC/ISI – Open Information Access • Michigan – Plan Coordination
Agent Domains DM • Group of agents registered with a common Domain Manager • Can be distributed across multiple hosts • Can be hierarchical/nested or “overlapping”(2002 implementation) • Provides for common administration and enforcement of domain-wide, VM-specific, and agent-specific policies.
UK Country 'Domain' Other Nation's 'Domain' US Country 'Domain' JFAC HQ's Organizational 'Domain' JTFHQ's Organizational 'Domain' Tactical Air Operations Functional 'Domain' Coalition Domain Types
Policy Concerns Go Beyond Typical Security Issues • Access management: Who can access what services? • Registration management: Who can join the domain under what circumstances? • Resource management: Who can have which kind and how much of a given computing resource? • Mobility management: What constraints should be placed on mobile code? • Communication management: What constraints govern conversations among agents? • Obligation management: Are agents meeting their commitments?
New Development: KAoS Agent Registration Helper • Generic KAoSGuard is instantiated automatically • Registration of Guard with Domain Manager handled • Guard enforces policies or delegates to other enforcers • Expanding set of policies that generic Guard can enforce • Domain Registration is handled automatically • KAoSAgentRegistrationHelper registers both with the Grid and with the Domain Manager • Domain Manager verifies that agent is guarded • Domain-aware agents will be able to: • register simultaneously with multiple domains • register and unregister at will • Collaboration with GITI resulted in many improvements in the latest CoABS Grid Release
Next Steps: KAoS Policy Representation (KPR) • DAML (DARPA Agent Markup Language) extends *ML semantics: • XML allows users to add arbitrary structures • RDF extends XML, meaning expressed in subject-verb-object triples • DAML extends RDF to express taxonomies and inference rules • Is being defined by DARPA and the W3C (www.daml.org) • KPR Description • DAML-based ontologies • Distinguishes between authorizations and obligations • Provides basis for: • Policy conflict resolution • Policy disclosure management • Reasoning about actions • Policy specification
NOMADS Mobility and Security • NOMADS adds mobility services to the CoABS Grid • Allows new agents to be launched to systems as needed • Allows agents to move across systems • Mobility may be forced (external event driven) or autonomous • Forced mobility scenario – Scramming agents from J-STARS to Ground Platform (18-month demo) • NOMADS provides secure execution of agents • Custom Aroma Java Virtual Machine provides extensive resource control capabilities • Enforces resource policies specified through KAoS • Protects against Denial-of-service attacks (shown in 9-month demo) • Dynamically adjusts resources based on mission priorities (30-month demo) • Working with Sun Labs to transition resource control into commercial Java
Technical Details • UWF/IHMC – Domain Management and Mobility & Security Services • Edinburgh/AIAI – Process and Event Panels • QinetiQ – Shared Visualisations and Data Translation • OBJS – eGents Agents Over E-mail • USC/ISI – Open Information Access • Michigan – Plan Coordination
Aim is a Workflow “Catch All” • Can take ANY requirement to: • Handle an issue • Perform an activity • Respect a constraint • Deals with these via: • Internal capabilities • External capabilities • Manual activity • Reroute or delegate to other panels or agents • Plan and execute a composite of these capabilities • Receives reports and interprets them to: • Understand current status of issues, activities and constraints • Understand current world state, especially status of process products • Help user control the situation • Copes with partial knowledge
I-X Approach • The I-X approach involves the use of shared models for task directed communication between human and computer agents who are jointly exploring (via some process(es)) a range of alternative options for the synthesis of an artifact such as a design or a plan (termed a product). • I-X system or agent has two cycles: • Handle Issues • Respect Domain Constraints • I-X system or agent carries out a (perhaps dynamically determined) process which leads to the production of (one or more alternative options for) a synthesised artifact. • I-X system or agent views the synthesised artifact as being represented by a set of constraints on the space of all possible artifacts in the domain.
DARPA CoABS Grid I-TEST, I-LEED & I-DEEL
Technical Details • UWF/IHMC – Domain Management and Mobility & Security Services • Edinburgh/AIAI – Process and Event Panels • QinetiQ – Shared Visualisations and Data Translation • OBJS – eGents Agents Over E-mail • USC/ISI – Open Information Access • Michigan – Plan Coordination
QinetiQ Technical Input • Storyboard & scenario data • Systems integration lead • Master Battle Planner • Flexible Visualisation • Use of Embodied Interface Agents / NLP • Translation and forwarding agents • Semantic Web Technology investigation
Flexible Visualisation • Allow customisable visualisation of any coalition data • Map-based visualisation using OpenMap (BBN Technologies) • Multiple projections; terrain data etc • Layers allow clean separation of information and rapid addition of new requirements • Agent-enabled for distributed visualisation of dynamic data • Data translation allows almost any XML input...
Translation Agents • Agent using XSL Transformations allows complete reformatting of XML message • Input via Grid (active or passive), file, HTTP... • Automatic translation and forwarding once set up (to any number of agents). • Rapid integration and change possible due to declarative mapping • Result: general-purpose “glue” for agent integration
Grid Other Agents HTTP Files Translation Agents Stylesheet <?xml version="1.0"?> <xsl:stylesheet xmlns:xsl= "http://www.w3.org/ 1999/XSL/Transform" version="1.0"> <xsl:output method="xml" indent="yes"/> <xsl:template match="obmsg"> <MBPDATA> <xsl:apply-templates select= "obs|obsvr"/> </MBPDATA> </xsl:template>
Translation Agents <agent_output date='Tue Jul 17 13:52:24 GMT+01:00 2001'> <OBJECT> <ROW> <hname>Big_Bill</hname> <location>34.51.80E/17.01.70N</location> <altitude>270</altitude> <velocity>5</velocity> <airtemp>91</airtemp> <humidity>31</humidity> <bodytemp>99</bodytemp> <bloodpressure>110/50</bloodpressure> <pulse>72</pulse> <basalskinresponse>430</basalskinresponse> <time>2012/09/24_17:00:00</time> </ROW></OBJECT> </agent_output> <?xml version="1.0" encoding="UTF-8"?> <MBPDATA> <MBPOBJECT> <GROUND> <ID>Elephant_Big_Bill</ID> <NOTES>Elephant data from Egents</NOTES> <LOCATION EW="E" NS="N" LON="34.8722" LAT="17.0361"/> <ALLEGIANCE VALUE="3"/> <STATUS VALUE="2"/> <DESCRIPTION>Elephant Herd Leader</DESCRIPTION> <MOBILITY VALUE="2"/> </GROUND> </MBPOBJECT> </MBPDATA>
Semantic Web • Investigations into Resource Description Framework (RDF), DARPA Agent Markup Language/Ontology Inference Layer (DAML+OIL) and the W3C Semantic Web initiative • “The way the web was meant to be” • Powerful and flexible storage and search of data - beyond databases • Potential for automated discovery and invocation of agents and services • Publication in SWWS
Semantic Web - Examples • Backing knowledge base for SitViewer • Auto translation (unique names; URIs) • ACL (more flexible than XML) • Assembly of partial data, with timestamp, origin and other metadata • Markup of human-readable documents