ARLINGTON, Va. - U.S. military researchers are asking for industry's help in improving machine autonomy technology sufficiently to enable its use in safety-critical applications, such as unmanned aircraft operating side-by-side with passenger planes in controlled airspace.
Officials of the U.S. Defense Advanced Research Projects Agency (DARPA) in Arlington, Va., have issued a broad agency announcement (HR001117S0045) for the Assured Autonomy project to assure that systems will operate safely and perform as expected, which will promote trust in machine autonomy and speed its adoption.
The goal of the Assured Auton-omy program is to develop rigorous design and analysis technologies to guarantee safety of autonomous machines that can learn on their own, based on experience.
This project will center on military autonomous vehicles. It will produce a set of publicly available software tools for use in commercial and defense sectors. DARPA seeks innovative techniques that render the learning algorithms inherently safe by incorporating safety constraints in the learning process, while meeting learning objectives.
Autonomous systems increasingly are critical to the military, and researchers have made tremendous advances in the last decade - particularly for unmanned vehicles that operate in the air, on the ground, and in the ocean.
Enabling these advances in machine autonomy have been innovations in sensor and actuator technologies; computing technologies; control theory; design methods and tools; and modeling and simulation technologies.
Despite these many advances in machine autonomy, however, adoption of such systems in safety-critical Department of Defense (DOD) applications remains to be both challenging and controversial. Designing in reliability to ensure trust is key to widespread use of machine autonomy.
The Assured Autonomy program has four technology areas (TAs): design for assurance; assurance monitoring and control; dynamic assurance; and integration and an experimentation platform. Proposers may address any of the four TAs.
Design for assurance involves a learning-enabled, cyber physical system (LE-CPS); assurance monitoring and control involves breakthroughs in operation-time techniques for assured and safe operation of LE-CPS for assurance monitoring and control; dynamic assurance involves new approaches in dynamic assurance that integrates design time assurance with operation time assurance; and the experimentation platform will provide challenge problems for evaluation and demonstration of the assurance technologies.
The program has three phases: an 18-month first phase for initial research and tool development; a 15-month second phase to focus on technology enhancement; and a 15-month third phase for technology maturation and demonstration on experimentation platforms.
Goals of the program are to increase scalability of design-time assurance; reduce overhead of operation-time assurance; scale up dynamic assurance; and reduce trials to assurance.
DARPA officials say they expect to make several contract awards for the Assured Autonomy program.
More information is online at www.fbo.gov/spg/ODA/DARPA/CMO/HR001117S0045/listing.html.