SBIR-STTR Award

We propose to demonstrate the feasibility of a decision-theoretic planning system for air operations consisting of: 1) an Asset and Payload Scheduling component to schedule assets and payload by taking into account missions at hand and assessed situation and threat using constraint programming technology; 2) a Proactive Flight Planning component to perform route planning for a mission based on multiple way points and the current situation and threat utilizing our in-house route planning algorithm successfully applied in previous projects involving course of action forecasting (Army Geospatial Center) and safe route planning (Office of Naval Research); and 3) a Dynamic Plan Adaptation component to dynamically adjust routes during a mission based on real-time situation and threat assessment using Partially Observable Markov Decision Process (POMDP) that implements a reward scheme where the most reward is obtained upon a successful strike. POMDP is suitable for handling stochastic transitions and observations in line with the uncertainty of moving a weapon delivery platform from one point to the other under variable threat. We plan to build a limited scope prototype for decision-theoretic planning for ATO objectives.

Benefit:
The software product to be developed in this effort addresses a fundamental requirement common all air operations, that is the need to effectively operate in a multidimensional operating environment involving multi-mission, multi-target, and multi-threat situations requiring the ability to dynamically re-plan missions and determine the most effective routes in flight. The solution to be developed can be applied to already existing frameworks and systems, such as JMPS and CMARS, and will be used by military planners and pilots to reduce cognitive demands and support decision making in the execution of ATOs.

Keywords:
Air Tasking Orders, Air Tasking Orders, dynamic plan adaptation, Partially Observable Markov Decision Process, proactive flight planning, decision theoretic planning

Unmanned Aircraft Systems (UAS) are being transformed from stand-alone assets to nodes in a network of interconnected knowledgeable entities in the operating environment working to enhance decision-making and collaboration in intelligence collection and precision strike missions. There is a well-established need for dynamic re-planning during a plan execution in response to environmental changes, UAS status, and mission updates. In Phase I, we designed and demonstrated the feasibility of a Decision-Theoretic Planning System for Air Operations, with our architecture consisting of three main components: Asset and Payload Scheduling that schedules assets and payload by taking into account the missions at hand and the assessed situation and threat analysis obtained using Constraint Programming (CP) technology; Proactive Flight Planning that performs route planning for a mission based on multiple way points and the current situation and threat; and Dynamic Plan Adaptation that dynamically adjusts routes during a mission based on real-time situation and threat assessment, and implements a reward scheme where the most reward is obtained upon a successful mission completion. In Phase II, we propose to research and develop a full-scale prototype for decision-theoretic planning/replanning system (dPLAN) for ATOs, and evaluate the performance of the developed system for UAS intelligence gathering missions.

Benefit:
The software product to be developed in this effort addresses a fundamental need common all air operations, that is the need to effectively operate in a multidimensional operating environment involving multi-mission, multi-target, and multi-threat situations requiring the ability to dynamically re-plan missions and determine the most effective routes in flight. The solution to be developed will be applied to already existing systems, such as JMPS-E, and CCS.

Keywords:
unmanned aircraft system, Partially Observable Markov Decision Process, efficient frontier, Constraint Propagation, proactive flight planning, decision theoretic planning, dynamic plan adaptation, Air Tasking Orders