The working group on knowledge-based planning systems for coalition command and control met on May 10-11, 1999 to discuss research. The working group comprised 23 participants from 4 nations (13 US, 6 UK, 1 Australian, 3 Canadian) and from several perspectives, including operational, technical, and program management). The group focused on identifying operational needs for coalition C2, technology needed to satisfy those needs, and research required to develop the technology.

Results briefed to the entire group focused on issues raised during the discussion, general themes identified by the group, long-term challenges, near-term opportunities, and recommended next steps. These results are summarized in this report.

The participants identified several issues faced by coalitions that are not generally a concern for unilateral operations. The principal themes addressed by the working group came from these issues which included:

• The need to develop shared perspectives among participants with dissimilar backgrounds, training, procedures and language. The need to come to agreement on objectives, perceptions, mechanisms, progress.
• Significantly increased operational complexity and the need to deal with mixes of activities involving bilateral and multilateral agreements, military coalitions, mixed civilian-military operations (Figure 1 shows an example of an array of participants and agreements example taken from a recent representative exercise)
• Less predictability for operations due to less opportunity to train and rehearse together
• Difficulties in creating and managing processes involving coalition partners—the more integrated, the more challenging
• Challenges in initiating knowledge-based operations, in particular difficulties and costs associated with knowledge acquisition
• Difficulty in moving quickly due to increased operational friction

The group acknowledged that political realities rather than functional requirements typically define coalitions. Coordination in such cases may be aimed at deconfliction rather than integration of operations. Yet it was our belief that integrated operations if accomplished effectively would yield significant benefits and our deliberations addressed issues associated with highly integrated activities.

The group considered areas where knowledge-based tools could assist in forming and managing operations. Specifically, the group considered capabilities that could be provided by knowledge-based tools that would assist coalition operational staffs. Furthermore, although the focus of the group is on planning systems, the group felt that execution management and replanning should be addressed as well. Identifying tools implementing these capabilities led to several themes for focusing further discussion emerged from the identified issues. These included:

• Tools to assist in developing shared perspectives
• Support for coalition formation and interaction
• Support for human interaction
• Shared plans and processes
• Assistance in managing coalition operations

These themes are elaborated further below.

Coalition members will typically have different perspectives of the battlespace, as well as different interests and objectives. Software that enabled exploration (say through simulation) of different alternatives could assist in developing a shared set of objectives. Tools to portray the physical and operational environment could assist in reaching agreement about methods for achieving objectives. Knowledge-based planning tools that elaborated intended operations could help participants to understand their roles and the interdependencies among activities.

Knowledge-based tools that enable analysis of potential interactions among coalition partners could assist in the formation of teams and help avoid future difficulties. For example, coalition members may have religious beliefs, dietary restrictions, and living standards that differ significantly and that could lead to internal friction. Software that helped to anticipate these difficulties would assist those responsible for forming, structuring and deploying coalitions to avoid potential difficulties.

Coalition members are likely to bring a variety of languages and different levels of technical sophistication to an operation. Multilingual system interfaces will ultimately be necessary to provide effective technical support. Machine aided translation will help in sharing information.

Policies for releasing information to coalition members will differ based both on individuals’ functional roles and national origin. Intelligent systems that manage and protect the flow of information will be necessary for smooth and proper functioning of coalition operations.

To ensure the success of integrated operations, all participants must share an understanding of the activities comprising the operation and their temporal, spatial, and functional relations to one another. Tools to manage centralized plans, keep them up to date, facilitate authoring by different planners, and disseminate changes will be essential to maintaining a common, consistent shared view and to ensuring that participant’s responsibilities are understood and met.

Intelligent process management tools that extend current workflow management concepts will greatly aid in ensuring that complex operations are carried out efficiently and correctly. Key advantages that could be provided by such tools include assistance in determining that task preconditions will be met and that resource requirements will be satisfied.

In discussing the themes outlined above, a number of desirable capabilities were identified. In general, technology does not currently exist to support the development of these capabilities. Advances in technology necessary to provide the capabilities will require focused programs of research and development spanning a range from basic research to experimental development. In this section, we discuss specific research topics and particular recommended areas of focus. The topics are loosely grouped into the following categories:

• Planning methods
• Information representation
• Execution support
• Learning
• Information infrastructure
• Modeling and knowledge acquisition

A great deal of work has gone into the development of techniques for automatically creating plans from a specification of objectives and operations or tasks that can change the world in specified and predictable ways. Most of these techniques are still inadequate to the task of planning in complex, dynamic, and uncertain worlds. Although funding agencies such as DARPA have begun addressing these challenges, the group felt that significant additional challenges remain. Specific areas that are felt to require attention include:

• Continuous planning methods that integrate incremental plan development with execution

• Constraint management—understanding how the space of allowable plans is changed by the incremental imposition of constraints

• Mixed initiative techniques— methods that share planning tasks among humans and machines according to their individual knowledge and capabilities

• Multiple objective planning—techniques to allow the achievement of multiple, possibly competing objectives
• Adversarial planning support that explicitly model opponents, their objectives, intentions, and beliefs
• Plan recognition methods for understanding from observations of plans what their objectives are
• Plan sketches and non-top-down methods that enable planners to fill in plan components as necessary and appropriate information becomes available

More extensive representations for information needed for planning are necessary if planning systems are going to be able to handle the challenging problems posed by coalition planning. Information that will need to be captured for efficient use includes

• Descriptions of plans, processes, organization
• Representations of authority, rules-of-engagement, legal restrictions
• Knowledge used for negotiation
• Uncertainty, temporal data, utilities
• Design rationale

Planning is only a part of successful operations. Support is also required for the execution of plans. In particular, capabilities for defining and integrating monitoring activities into plans, and using those monitoring activities to assess the current needs and likelihood of success of plans will be required for smooth, successful execution. The development and integration of focused, faster-than-real-time simulation into plan execution can provide a kind of feed-forward control capability.

Intelligent workflow management capabilities that can smoothly integrate planning and execution will need to be developed.

The ability to operate successfully in situations where information is missing, inconsistent or of reduced quality will be necessary. This will require significant progress beyond current logic-based and hard-constraint-based planning methods.

It will be important to capture and use lessons learned from experience. Better learning capabilities will be necessary to build knowledge bases quickly and correctly.

There is significant investment and reliance on legacy databases and systems and it will be necessary to use these legacies into future intelligent systems. This will entail sophisticated wrappers and interfaces for these systems that provide not only information about how to access the systems, but also descriptions of the capabilities and requirements of the tools for use by planning and reasoning agents.

Intelligent command-and-control systems will be voracious consumers of information. To ensure continuous and adequate development and dissemination of information, intelligent information management capabilities will need to be developed. Capabilities to provide information to uses without requiring significant assistance from the user in defining and locating information sources will be required. The ability to push important information to the user without being asked will also be necessary to avoid unpleasant surprises.

The infrastructure must incorporate security capabilities that can help manage flows of information and ensure that only authorized users (or agents) receive sensitive information. Security measures should not add significant overhead to ordinary processing if the overall operational capability is not to be compromised.

Knowledge acquisition and modeling capabilities will need to be significantly improved if knowledge-based systems are to be constructed quickly. Users should be able to develop their own knowledge bases and modify existing knowledge bases to adapt to new situations.

Models of participants’ activities and objectives will be necessary to quickly create systems for planning and monitoring complex, multiplayer, coalition activities. Models of partners’, opponents’, and neutral players’ roles, capabilities and goals will be necessary for developing operational plans.

The working group participants agreed that new, long-term research programs were needed to develop the capabilities identified in our discussions. At the same time, however, the group felt that initial steps towards developing intelligent capabilities for command and control could be taken by building on ongoing activities. Specific starting points are offered by the DARPA Joint Task Force ATD program and the Adaptive Course-of-Action ACTD carried out at the Pacific Command (PACOM) by the DISA AITS Joint Program Office. The group felt that the Collaborative Virtual Workspace (CVW) tool, in particular, could provide a low-cost coalition collaboration environment and thus offer a means to jumpstart a development.

The group recommended a phased set of demonstrations built around a coalition exercise.

• Phase 0 would provide CVW to members of a coalition exercise and use it to support the joint development of shared objectives and plans (starting immediately)
• Phase 1 would add newly developed ACOA tools to the framework and enable participants to capture and monitor developed plans (approximately one year)
• Phase 2 would introduce additional knowledge-based tools to support (approximately 1-2 years)

• Constraint management
• Representations
• Process management

The group felt that the phases described above could be carried out in the context of a coalition experiment (COEX) as proposed by other working groups. The group also recommended that the research suggestions be synthesized into a strawman program plan for elaboration and discussion with the appropriate funding agencies. The group felt that it was critical to build a coalition to support both the near term and long term recommendations.

A working group composed of representatives of several nations and spanning a range of operational, program management, and technical experts met to discuss issues the knowledge-based technology needs posed by the challenges of modern, dynamic command-and-control in a coalition environment. A number of important issues raised specifically by coalition operations were identified and these led to themes based on desired capabilities that were addressed by the group. These themes then led to a discussion of the research required in order to be able to develop the necessary capabilities. The research recommendations are outlined above. The group finally considered near term objectives to start a program to develop knowledge-based tools for coalition command and control. The group identified a phased set of low-cost practical experiments and our recommendations are outlined above.