ARLINGTON, Va. – U.S. military researchers are asking business to develop dependable and trusted artificial intelligence (AI), machine studying, and programs autonomy applied sciences for future command and control, surveillance, and logistics programs.
Officials of the U.S. Defense Advanced Research Projects Agency (DARPA) in Arlington, Va., issued a broad company announcement (HR001122S0039) final week for the Assured Neuro Symbolic Learning and Reasoning (ANSR) mission.
The ANSR program seeks breakthrough improvements in new hybrid AI algorithms for military command and control that deeply combine symbolic reasoning with data-driven studying to create strong, assured, and reliable programs.
Researchers say hybrid AI algorithms in a position to purchase and combine symbolic information, and to carry out symbolic reasoning at scale, will ship strong inference, generalize to new conditions, and present proof for assurance and belief.
Related: Air Force picks two corporations to develop mission planning autonomy to facilitate manned-unmanned teaming
Machine autonomy must be to hasten the tempo of military operations, cut back cognitive calls for on the warfighter, and enhance standoff distances to assist preserve warfighters secure, researchers say.
The prevailing development in industrial machine studying analysis is in direction of scaling as much as giga- and tera-scale fashions with tons of of billions of parameters to enhance accuracies and efficiency.
These tendencies are usually not sustainable, nevertheless, due to the extraordinarily excessive computational and knowledge wants for coaching such fashions, and the shortcoming of data- and compute-starved military functions to entry cloud-scale pc assets.
U.S. military functions, as well as, are security and mission-critical, and so must function in unseen environments, have to be auditable, and have to be trustable by human operators. Instead, the ANSR program seeks to develop new hybrid AI algorithms which might be extra conducive for use in real-world military functions.
Related: Military researchers ask business for superior microelectronics for machine autonomy and sensor processing
ANSR seeks to switch coaching and inference procedures to interleave symbolic and neural representations that can lead to representations which might be grounded in domain-specific symbols, and can embody prior information and domain-specific guidelines.
The program will display assured execution of an unaided intelligence, surveillance, and reconnaissance (ISR) mission to develop a standard working image of a dense city atmosphere. The autonomous system will carry an results payload to scale back sensor- to-effects supply time.
The capabilities required of the autonomous system immediately are usually not achievable by state-of-the-art machine studying or standalone symbolic reasoning programs. The ANSR program has 4 technical areas: algorithms and structure; specification and assurance; platforms and functionality demonstration; and assurance assays and analysis.
Related: DARPA to outfit F-16D jet fighter with artificial intelligence (AI) to spice up belief in AI as a human accomplice
The guiding problem for this system would be the assured execution of an unaided ISR mission in a dynamic dense city atmosphere.
Algorithms and structure seeks to develop and mannequin new AI algorithms and architectures that deeply combine symbolic reasoning with data-driven machine studying. Specification and assurance will develop an assurance framework and strategies for deriving and integrating proof of correctness and quantifying mission-specific dangers.
Platforms and functionality demonstration will develop use-cases for engineering functions of hybrid AI algorithms for strong and assured efficiency. Assurance assays and analysis will develop an assurance check harness with adversarial AI, and consider these applied sciences in programs.
Related: Air Force researchers eye methods to allow autonomous unmanned plane operations in managed airspace
The program will search to display an autonomous ISR mission that requires the warfighter solely to determine the realm of curiosity, and allow AI to investigate the wants for navigation and actions towards the threats.
Companies ought to add abstracts no later than 13 June 2022 and proposals no later than 26 July 2022 to the DARPA BAA Website at https://baa.darpa.mil.
Email questions or issues to Sandeep Neema, the ANSR program supervisor, at [email protected]. More data is on-line at https://sam.gov/opp/0c28fb55fcb446dc95ed3337b385b36c/view.
https://www.militaryaerospace.com/computers/article/14277721/artificial-intelligence-ai-machine-autonomy-command-and-control
