Proceedings of the 81.2 Military Operations Research Society (MORS) Symposium, Alexandria, VA, (June 2013)
The dependability, persistence, and versatility of unmanned aerial systems (UAS) have made them indispensable assets for providing intelligence, surveillance, and reconnaissance (ISR) over the battlefield. As larger constellations of heterogeneous, multi-purpose UAS are tasked to perform more diverse missions in unpredictable, dynamic environments with more sophisticated sensors, they are transitioning from remote control into the realm of autonomy. Intelligent systems have the potential to augment human capabilities in collection planning for these missions, but it is essential to understand how they are able to optimize against multiple competing objectives.
Generating optimal collection plans for airborne ISR assets remains a challenging problem given the complexity of the search space, the number and variety of aircraft available to perform these missions, and the complexity of the underlying geospatial data available. Not only does a collection plan have to satisfy multiple constraints such as collision avoidance, it must also meet numerous and often competing objectives such as avoiding adverse weather effects and maximizing sensor performance. These challenges, coupled with the increasing number of sensor types and aircraft platforms, and the highly diverse advantages and limitations of each, present a challenge for ISR collection planners and UAS operators to optimize their use. For mission planners to develop confidence in semi-autonomous mission planning decision aids, and to use them effectively, it is necessary demonstrate the ability to reliably accomplish mission objectives by providing analysts, planners, and operators with a deep understanding of the capabilities and limitations of their aircraft during varying mission scenarios.
Our approach to support the understanding of ISR effectiveness is based on our SPARTEN (Spatially Produced Airspace Routes from Tactical Evolved Networks) tactical decision aid that generates coordinated mission plans for UAS constellations by allowing the mission planner to specify which objectives are important to them for each mission based on their specific tactical needs. We have formalized the computation of specific fitness values for each of the optimization objectives and constraints under consideration. The constraints are collision avoidance and flight endurance, and the objectives are adverse weather avoidance, area of analysis (AOA) coverage, ground control station (GCS) range, latency, linearity, military value of terrain, named area of interest (NAI) emphasis, restricted operating zone (ROZ) compliance, acoustic footprint, and sensor performance.
The nature of these objectives allow a priori computation of local fitness values in time and space using cost factor maps (CFMs) that are attributed on the underlying air maneuver network (AMN). The AMN is a topological representation of the operational airspace containing a planar network whose nodes represent air control points and launch sites for UAS and are connected by edges. The AMN maintains topological and temporal reference frames to represent flyable air corridors and provides a data structure that supports treating continuous airspace as a discrete state space, where a valid route for any asset is an ordered list of edges that starts and ends at a launch site.
CFMs are concise representations of the mission plan tradeoff spaces that facilitate the rapid computation of fitness values for candidate collection plans. The CFMs are stored as numeric attributes for each AMN edge calculated using the fitness value formulae and the underlying time-varying geospatial data. Each CFM contains geospatially specified data for each edge on the AMN that indicates how that region in space and time is affected by the given concern. By computing the CFM in advance, we can offload much of the computational complexity involved in fitness evaluation so it can be done quickly. So, each AMN edge carries a list of attributes relating to flight along that edge for each time interval for each optimization objective.
We have tested our approach using geospatial data from the Caspian Sea and Afghanistan areas of operation by generating coordinated mission flight plans in NATO’s Common Route Definition (CRD) format. Solution sets include geographic information system (GIS) visualization of solution metric fields using ESRI’s Commercial Joint Mapping Toolkit (CJMTK) to highlight the strengths and weaknesses of each collection plan as viewed through the ‘lens’ of each optimization objective.
We present a method to understand and evaluate collection plans for constellations of UAS and provided visualization aids that help operational users make sense of the results. Our approach provides a promising decision aid to help mission planners understand the effectiveness of their mission plans for constellations of UAS by making use of available geospatial and temporal data. Our computational formulae for each constraint and optimization objective successfully represent quantifications of complex time-varying geospatial data and summarize the efficacy of a given ISR collection plan for multiple heterogeneous UAS assets. This enables the generation of useful collection routes in near real time based on rigorous mathematical foundations while also reflecting the operational needs of the warfighter for specific missions.
By employing novel visualization techniques using GIS to represent their effectiveness, we help the user ‘look under the hood’ of the collection planning algorithm and understand the viability and effectiveness of mission plans to identify coverage gaps and other inefficiencies.
1 Charles River Analytics
2 U.S. Army Construction Engineering Research Laboratory
For More Information
(Please include your name, address, organization, and the paper reference. Requests without this information will not be honored.)