STO-Activities: (no title)

Activity title: AI tools for Operational Planning
Activity Reference: SAS-200
Panel: SAS
Security Classification: NATO UNCLASSIFIED
Status: Proposed
Activity type: RTG
Start date: 2024-09-01T00:00:00Z
Actual End date: 2027-09-01T00:00:00Z
Keywords: Artificial intelligence, Decision Making, Decision support, Deductive reasoning, Machine learning, Operational planning
Background: At its heart, machine learning (ML) is a sophisticated pattern-recognition system trained by an algorithm. Advances in ML offer promising ways to help analysts in command and control find new insights (patterns from data) necessary for decision making, developing knowledge, enabling understanding, and producing predictive assessments. However, developing knowledge and enabling understanding can also be obtained using a deductive reasoning approach, namely architectures that use formal language for knowledge representation and have abilities to reason about novel concepts or tasks, from example. In this proposal, we will classify Artificial Intelligence (AI) tools in two categories: AI/ML for solving unique tasks (Statistical-Pattern-based) and Reasoning AI (Logic-Rules-based). The role of AI in military systems is one of the most important considerations for defense policy makers for the near future. Since AI can predict the outcome of a decision faster than a human, it can bring an undeniable advantage to warfighting: when a pilot needs to choose amongst speed, altitude, pitch, yaw, weapon selection, enemy relative bearing, and terrain avoidance, an AI system can easily exceed the ability of a human to observe, orientate, decide and act. Consequently, the development of AI tools to support tactical actions, such as destroying enemy forces and navigating from one point to another, provides a potential tactical advantage. However, tactical actions are of limited scope and duration; integrating AI only at the tactical level disregards the decisive effects that take place at the operational level. Operational level warfare is concerned with the preparation and execution of campaigns and major operations, across one or multiple domains (MDO). Operational planning handles primarily the what questions: what is the end-state, what objectives will achieve it, what effects must be created to achieve the objectives, and what tasks and action will produce those effects. The tactical planning level is concerned with how to achieve assigned missions and objectives using the resources provided. The North Atlantic Treaty Organization’s (NATO) AJP-5 Allied Joint Doctrine for the Planning of Operations describes the Operations Planning Process (OPP) from the initiation phase, through to the orientation, design, plan development, approval and execution phases, and addresses plan review, revision and cancellation. The planning activities from the OPP are: (i) initiation; (ii) mission analysis; (iii) courses of action (COA) development, COA analysis, COA validation and comparison, and the commander’s COA decision; (iv) plan development; and (v) plan review. The rationale underlying the five stages of the OPP is found, to one degree or another, in Allied planning processes such as the NATO Comprehensive Operations Planning Directive (COPD), the US Army’s Military Decision Making Process (MDMP), the US Marine Corps’ Planning Process (MCPP), the Canadian CFJP 5-0 Operational Planning Process, the British Army’s Tactical Estimate (TE), the French Méthode d’élaboration d’une décision opérationnelle (MEDO), and those of other NATO nations.
Objectives: The objective of this RTG is to investigate how AI may be best applied in Operational Planning to support defence and security decision making within NATO, its member Nations, and partners. The goal is not to explore tools that will replace military planners, but rather that will increase their productivity and enhance the plans that they produce. In order to achieve this goal, it is proposed that the RTG will undertake six tasks:
Task 1: Develop a synopsis of current AI technological developments related to OPP requirements.
Task 1. 1 Compare and contrast existing national AI efforts in the following J-staff and staff coordination functions:
J1-Personnel [e.g. Operational Staffing Documents];
J2-Intelligence [e.g. PED, Image/Text Analysis, intelligence support to targeting];
J3-Operations [e.g. Plan-following, Synch-Matrix automation, LOOs-Decision Point Logic and Status tracking];
J4-Logistics [e.g. Supply chain analysis and status awareness, Predictive maintenance];
J5-Plans: [e.g. Wargaming with intelligent agents red-teaming and Courses of Action testing, Dynamic Retasking of Assets, Mission Rehearsal];
J6-Communication Systems (e.g. protection from known or anticipated threats); Information Activities Coordination Board (e.g. Information Environment status or Information Environment effects generated from planned operations).
J9 – Civil Environment Assessment
Task 1. 2 Follow/track the development of the TOPFAS (Tool for Operational Planning, Force Activation and Simulation) AI Assistant developed by NCI Agency (sponsored by ACT and SHAPE).
Task 2: Scoping feasibility of distinguished AI technologies.
Task 2.1 Provide a very concise overview of potential AI technologies feasible for operational planning: Machine Learning (Statistical-Pattern-based), Symbolic AI (Reasoning with Logic and Rules), and Hybrid approaches. Report the benefits and drawbacks of each category.
Task 2.2 Investigate the benefits of knowledge representations for solving the planning problems. Explore the value of developing an ontology that can be used to describe/represent the elements of a planning domain, namely relations among objects of a planning domain, as well as relations among actions, tasks, plans, and goals.
Task 2. 3 Investigate the relationship between application characteristics and AI technologies (i.e., a particular AI technology may be fruitful for application A, but not for application B).
Task 3: Prepare an inventory of opportunities for AI-support in Operational Planning.
Task 4: Research one opportunity selected from the above inventory.
Task 5: Develop a scenario as context to demonstrate the solution for the opportunity identified in Task 4.
The scenario should represent the planning process, as well as the AI support, to a functional level that allows for evaluation and feedback by military domain experts. The following subtasks will be covered:
• Task 5.1. Analyze operational planning data to identify potential problems and develop/propose solutions. [Make use of expertise in operational community].
• Task 5.2. Evaluate multiple options to identify the best Course of Action (COA) based on the desired outcomes and constraints.
• Task 5.3. Generate simulations using the scenario to evaluate potential plans.
• Task 5.4. Assist in resource allocation and task prioritization. [Identify (potential) sponsors for future work]
• Task 5.5. Assess the progress in reaching the mission end-state and objectives.
• Task 5.6. Assess the human operator pre-experience, learning curve, and remaining planning effort. Identify what improvements are made by adding AI support.
Task 6: Develop conclusions and recommendations.
Topics: With a particular focus on supporting defence and security decision making during the OPP, the scientific topics to be covered in the RTG are:
a) Developing a structured approach to classify, contrast, and compare ML/Statistical-Pattern-based, Reasoning AI, or both methods combined;
b) Identifying best practices from the current implementation of the AI tools with the aim to accelerate their adoption in more phases/processes during the OPP;
c) Exploring how computer-assisted and AI-augmented Operational Planning drastically reduce friction and the fog of war, decreasing the uncertainty and the disorder of the combat.
Contact:
Open2Partners:
Title:

Created at 04/03/2024 18:00 by System Account
Last modified at 16/05/2024 06:00 by System Account
 
Go back to list
Home(NATO STO)