On-board Planning System for UAVs Supporting Expeditionary Reconnaissance and Surveillance
Aurora and MIT are working on a program for ONR in which we are developing the architecture,
core algorithms, and human interface concepts for a multi-platform, distributed
UAV-USV team that responds to requests from field operators for intelligence support.
Aurora is building on existing robust distributed tasking algorithms that have been
demonstrated to work in intermittent communications environments, tailoring them
to address multiple-operator issues, rules of engagement constraints, and the necessity
to ensure performance of time-critical tasks.
Capabilities are being incorporated into a collaborative decision-making process
flow that specifically accounts for human supervisory control issues, including
interfaces, cognitive roles, and situational awareness. This system will be implemented
in an onboard planning module, already in development, that can be incorporated
into low cost UAVs, giving them higher levels of autonomy and making it possible
for them to coordinate their activities as a team over a real-world communication
Aurora is working with NASA to develop and demonstrate a method for generating robust
autonomous flare maneuvers for manned and unmanned vehicles. The goal of a flare
maneuver is to safely transition an aircraft from final approach to touchdown, decelerating
the vehicle and setting up a safe landing attitude.
During this transition, a complex series of dynamic events can occur, and the pilot
or autopilot must address uncertainty in aerodynamics, disturbances such as cross-winds
and, in severe circumstances, aircraft impairment. Aurora’s Flare Planner takes
advantage of recent advances in control theory which allow for fast ‘on-the-fly’
determination of appropriate control inputs for complex dynamic situations – these
methods efficiently generate flight paths that simultaneously satisfy vehicle performance
limits, constraints, and touchdown criteria.
Distributed Sensor Fusion
In the expanding arena of Net-Centric Warfare (NCW) there is still a capability
gap with tactical UAV employment: incorporation of local UAV data into the intelligence
datastream is still limited, and more importantly, coordination of data gathering
platforms (especially UAVs) is not automated or optimized. As the number of UAVs
in the battlespace increases, the potential for fast and accurate localization,
identification, designation, and prosecution of time-sensitive targets will only
be realized if mechanisms for coordinating assets are developed. This includes coordination
of multiple local, mixed assets, as well as combined local and stand-off assets.
The focus here is on managing UAV resources by appropriately tasking them to perform
‘coincident collection’ of ISR data, either with other UAVs – placing multiple UAV
sensors on a target to mitigate the sensor limitations of single vehicles – or with
stand-off assets – providing low-altitude data collection on targets identified
by stand-off sensors (GMTI, SAR) and thus increase the fidelity of identification
or provide target designation.
Under a Phase II SBIR sponsored by the Air Force Research Lab, Aurora is exploring
advanced algorithms to optimize the autonomous management of multiple UAVs to both
make UAV data collection more relevant to overall battlefield situational awareness,
as well as optimizing the coordination of multiple UAVs and their sensors to ensure
effective, timely, and persistent ISR information.