Using a Structured Plan Representation to Support Multilevel Planning

1
Using a Structured Plan Representation to Support
Multilevel Planning

• Michael C. Dorneich, David Mott, Jitu Patel,
  Edward Gentle

Knowledge Systems for Coalition Operations 1
April 2009, Chilworth Manor, Southampton, UK
2
Outline

• Introduction
• Planning Representations
• Collaborative Planning Model
• CPM-based Experimental Tools
• Evaluation
• Evaluation Objectives
• Participants
• Schedule and Logistics
• Scenario
• Findings

3
Premise

• Military planning must achieve the coordination
  of multiple agents towards a common goal
• Therefore planning depends on ....
• Communication of a common understanding of
  commander's intent,
• Communication of objectives, resources, and
  constraints.
• Understanding of the rationale for decisions made
  at any level of planning, in terms of the options
  chosen or alternatives rejected

4
Collaborative Planning Model (CPM)

• Planning is problem solving, making Assumptions
  and Decisions under Constraints
• Goals (What) are turned into Tasks (How)
• Problem solving is distributed across human
  agents via Collaborations
• Plans are explained and explored via Rationale
• Distributed agents need to use Teamwork
• Planning is performed in a Context
• These concepts are specified as a logical model
  in OWL

Visualisation
Rationale
Plan representation
English-like language
Digitised Semantics

• Concepts derived from
• I-N-O-V-A and SPAR (Tate et al)
• PLANET (Gill Blythe)
• FlyPAST (Mott Gadsden)
• ATMS (de Kleer)
• Joint Intention Theory (Levesque)
• Common Logic
• Controlled English (Sowa)
• US and UK Military Manuals
• extensions as part of ITA research

5
Collaboration

• Collaborations are shared problem solving
• Model the hand over of a Problem from setting
  Agent to solving Agent
• constraint-based partial Plan
• setter provides the Goals to be solved and the
  Constraints
• solver must solve within the Constraints
• collaboration is the link from high level Problem
  to the low level Plan solution
• Also need to model Synchronisations
• set of inter-problem Constraints coordinating
  peer to peer tasks (eg flying programme)

6
Visualiser (IBM)

• Graphical representation of the spatial and non
  spatial aspects of the plan
• Not intended as high quality User Interface!
• Edit plan objectives, tasks, resource requests,
  and assignments
• Import and export CPM in OWL and Controlled
  English
• Capture of rationale in Controlled English and
  graphical notation
• Calculate logical implications of the current
  plan, e.g. temporal constraints
• Calculate and visualise the rationale for plan
  items
• Calculate the effects of changing assumptions

7
PlanEditor (Honeywell)

• Graphical representation of the spatial and non
  spatial aspects of the plan
• Edit plan objectives, tasks, resource requests,
  and assignments
• Import and export CPM in OWL
• Display and capture of the rationale
• Automatic layout of plan entities

8
Evaluation Objectives

• Explore the representational power of the CPM in
  supporting multilevel collaborative planning
• Represent basic planning concepts
• Import and export plans
• Merge sub-plans and their rationale
• Support the detection of the reasons for
  conflicts
• Demonstrate integration possibilities of CPM
• Integration of multiple planning tools via CPM.
• What it is not
• Evaluation of the tools or their visualization
  capabilities

9
Protocol Scenario Overview
Create main plan
Merge plan
Finalize plan
IBM Visualiser
Subplan Task A
IBM Visualiser
Replan
Battalion Cmdr, Planner 1
Replan
Main plan Task A
Main plan Task B
Final Plan
Subplan Task B
Create sub-plan
Create sub-plan

• Participants
• Industry
• IBM UK
• Boeing USA
• Honeywell USA
• Government
• Dstl
• ARL

Company Cmdr, Planner 2A
Company Cmdr, Planner 2B
Honeywell PlanEditor
IBM Visualiser
Request Resources
Resource conflict
Detect resource conflict
IBM ArtyPro simulator
10
Brigade (X) and three Battalions

• Armoured Brigade to defeat hostile brigade
• Battalion 1 2 to build bridge and secure area
  after bridge
• Requires Engineering and Artillery resources to
  handle hostile force in south.
• Once finished, Battalion 3 will push through
  across bridges towards enemy,
• Requires Artillery Support to attack and fix
  hostile battalions.
• Feint to north to draw forces away from Battalion
  1 and 2s activities (not shown).
• Combat support of Close Support Artillery (owned
  by Brigade but on loan to each Battalion).

• Potential for Battalions 1 and 2 to both request
  use of Artillery at the same time
• targets in the North (at Brigade level) and
• targets in the south (at Battalion level),
• these might be conflicting as resources were
  limited.

11
The main Plan

• Planner 2A
• secure Nth LD for Armd Regt
• Clear up to the river
• Seize the River Crossing Site 1/2
• Defeat the Enemy in north
• Secure the north LD for the armed regiment.

• Planner 2B
• secure Sth LD for Armd Regt
• Clear up to the river
• Seize the River Crossing Site 1/2
• Defeat the Enemy in south
• Secure the south LD for the armed regiment.

12
Completed Sub-plan 2A
13
Completed Sub-plan 2B
14
Merging of Subplans

• Sub-plans imported into ARTYPRO tool to display
  resource requests on a common schedule chart
• Import includes some rationale for the plan
  entities, used to detect resource request
  conflicts.
• Final merge and visualization of sub-plans

Sub-plan 2A
Sub-plan 2B
15
Inconsistencies and Conflicts

• Inconsistencies across the subplans
• Detected post-evaluation after adding min
  durations for subplan 2A
• Calculation and display of key aspects of the
  plan that caused the conflict

16
CPM for multi-level planning

• Both planners able to capture the tasks in the
  plan
• Plans can be imported and exported between
  different tools created by different
  organizations via CPM/OWL
• basic plan entities
• rationale for these entities
• Multilevel planning can be achieved via problem
  solving collaborations working on different
  parts of the plan, based around hierarchical
  organizational structures.

17
Planning Issues

• Planner 2A wanted to violate some timing
  constraints imposed by main plan
• Looking to steal time build up time reserve
• Normally would clear with commander
• More detailed rationale would have been helpful
• Aspects of plan depend on variables unknown at
  the outset (2A, 2B)
• Decision points needed
• Planner 2B did not understand some aspects of
  main plan
• More detailed rationale is key

18
Planning Perspectives

• Planner 2A worked interchangeably in three
  different planning perspectives
• Task oriented view
• Creating tasks, ordering constraints
• Currently supported by tools and CPM
• Spatial view
• Location of forces
• Movement coordination
• Resource/timeline view
• Assigning performing resource
• Assigning reserves

19
Improving Representational Power

• Later version of CPM contains decision points to
  represent conditional tasks
• Later version of CPM clarifies tasks and
  objectives
• Need more complex logical definitions of planning
  concepts
• Future ITA work will address this, going beyond
  OWL

20
Rationale

• We hypothesise that rationale is key to shared
  understanding of plans
• Planner 2B articulated a constant stream of
  rationale, explaining what he was doing and why.
• is this an artifact of the experiment?
• We can capture it, represent it, and use it
• But how can this be captured within an
  operational context?

There is currently no bridge across the river a
bridge is needed and I assume that the bridge
will not be built in time, so cant use Armour.
Best remaining is Infantry. Therefore use Inf Coy
2/1. Since the reconnaissance unit is there, we
can use them too.
Given that this is line of sight, Armour can be
used. Armour is the strongest available resource.
We need strongest as the enemy MUST be removed.
Therefore use Armd Sqn 4
21
Summary

• Experimental tools were developed independently
  based upon the specification of CPM
• The evaluation demonstrated the integration
  possibilities of CPM in support of distributed,
  multi-level military planning.
• high level military planner developed a main plan
  and delegate the planning of sub-plans to two low
  level planners.
• two low level military planners developed the
  sub-plans in isolation
• the two sub-plans and their rationale were merged
  against a resource schedule chart to provide
  resources and check conflicts
• the sub-plans were merged with the final plan to
  provide a holistic solution to the main planning
  task
• The evaluation provided insights into the
  possible use of rationale and improvements
  required to the CPM and the planning tools
• The research and evaluation continues in the ITA
  programme

22
Questions?
23
Work In Progress