Challenges and Progress in Behavior-Based Adaptive Autonomy

Arjuna Balasuriya, Tyler Mayer, and Gerald Fry

AUVSI XPONENTIAL 2023, Denver, CO (May 2023)

Today’s uncrewed platforms are typically operated by humans using remote control to guide every detailed aspect of a mission. However, as missions become more complex, there are many scenarios (particularly in the marine, ground, and space domains) in which operators are unable to communicate with these un[1]crewed platforms in real time (due to adverse environmental conditions, regulatory restrictions on communications in ecologically sensitive areas, active interference by adversaries, or the desire to remain covert), making it challenging to recalibrate and update mission and control parameters on the fly. Fortunately, significant technical advances in onboard computing power and enhanced sensors offer a pathway to a level of autonomy that can overcome such communications limitations. Behavior-based autonomy architectures enable onboard mission autonomy software to select optimal behaviors for the mission at hand, assess the evolving state of the mission, and adapt to changing environmental conditions. Developers and operators of autonomy systems for challenging environments will gain insights into the importance of onboard behavior and parameter adaptation to the success of long-duration uncrewed platform missions in unpredictable situations and unknown environments. Designers will benefit from a discussion of how multi-objective optimization and reinforcement learning techniques are currently being used to enable assured autonomy onboard these uncrewed platforms.

For More Information

To learn more or request a copy of a paper (if available), contact Gerald Fry.

(Please include your name, address, organization, and the paper reference. Requests without this information will not be honored.)