La Unidad de Innovaci\u00f3n en Defensa (DIU) del Pent\u00e1gono (US-DoD) ofrece hasta US$ 100 Millones en futuros contratos, para que las empresas tecnol\u00f3gicas presenten en un plazo de 10 d\u00edas, propuestas de un nuevo tipo de software que emplee IA y al que denominaron \u201cOrchestrator\u201d.\u00a0El mismo debe permitir\u00a0que un solo operador pueda dirigir enjambres de drones en misiones complejas, mediante el empleo de \u00f3rdenes enviadas en formato de texto y voz. Las empresas preseleccionadas presentar\u00e1n sus prototipos en el plazo fijado (10 d\u00edas) para la primera evaluaci\u00f3n. Los oferentes que pasen esta etapa, continuar\u00e1n un per\u00edodo de 6 meses de evaluaciones sucesivas, que se realizar\u00e1n en todo tipo de plataformas aut\u00f3nomas terrestres, a\u00e9reas y marinas. Empresas especializadas en IA como Auterion y Primordial, estar\u00edan entre las interesadas y en capacidad de participar de este revolucionario proyecto.<\/p>\n
WASHINGTON \u2014 \u00a0The Pentagon\u2019s embassy in Silicon Valley, the Defense Innovation Unit (DIU), is offering\u00a0up to $100 million dollars in prizes<\/a>\u00a0for companies to prototype new user-friendly \u201corchestrator\u201d control\u00a0software for whole formations of unmanned air, ground, and water vehicles.<\/p>\n Submissions are due by Jan. 25 and, in keeping with\u00a0Secretary Pete Hegseth\u2019s need for speed<\/a>, the pace will be brutally brisk. \u201cIf selected, performers must be able to begin Sprint 1 testing within 10 days of selection notification,\u201d says the\u00a0detailed guidance<\/a>\u00a0for interested companies, with successively more complex stages following fast over the next six months.<\/p>\n Backed by the US Navy and the Defense Autonomous Warfare Group (DAWG,\u00a0carrying on work<\/a>\u00a0from the original Replicator initaitive), the project aims to move beyond the current, labor-intensive model in which each drone requires at least one trained human operator to constantly direct it by remote control during all but the most basic tasks. Instead, DIU wants to develop an\u00a0Autonomous Vehicle Orchestrator<\/a>, an AI mediator between human and machine that turns plain-English instructions into detailed machine-readable commands the robots can execute.<\/p>\n In theory that \u201corchestrator\u201d would allow a single ordinary servicemember, without special training, to give broad commands to a whole group of unmanned vehicles at once \u2014 by text or voice.<\/p>\n \u201cWe want orchestrator technologies that allow humans to work the way they already command \u2014 through plain language that expresses desired effects, constraints, timing, and priorities \u2014 not by clicking through menus or programming behaviors,\u201d said Marine Lt. Gen. Frank Donovan, the three-star director of the DAWG, in the\u00a0official announcement<\/a>\u00a0published Tuesday.<\/p>\n Proposed prototypes must be able to understand and execute common military commands, the\u00a0guidance for industry<\/a>\u00a0explains, like \u201cplace crafts 1-5 in echelon left\u201d or \u201chold position, conserve battery, and wait for further tasking unless a threat crosses Line Bravo.\u201d<\/p>\n Voice control is probably the project\u2019s biggest challenge: If you think refilling prescriptions or making reservations with an AI voice can get frustrating, imagine trying to win a war with one. While the explosion of Large Language Models has revolutionized AI\u2019s ability to communicate with humans in plain text, voice communication still lags behind.<\/p>\n But most people talk faster than they type, especially under stress, and in combat, troops want to have their hands on a weapon, not on a keyboard, and their eyes scanning for danger, not staring at a screen. So there\u2019s real value in voice control that can work even amidst the roar and terror of a battlefield. At least one company,\u00a0Primordial<\/a>,\u00a0says it has field-tested voice-operated drones for\u00a0Army special operators<\/a>\u00a0(and for reporters, including from Breaking Defense). The DIU\u2019s $100 million in prize money should bring other startups out of stealth mode.<\/p>\n The other complex technical challenge is likely to be found in making the \u201corchestrator\u201d AI smart enough to turn the human\u2019s voice or text commands into a detailed plan coordinating multiple drones, potentially of different types \u2014 say, drones flying cover for robot boats or tanks \u2014 and then adapt that plan as it acquires new information, without freezing up and asking the human for new instructions. Indeed, the submission guidelines make clear that the AI might not always be able to reach a human because enemy jamming or natural obstacles might block transmissions.<\/p>\n \u201cThe Orchestrator must be designed to function effectively under intermittent connectivity. The Orchestrator may also eventually need to be run in a disconnected, edge environment without access to the cloud,\u201d the\u00a0guidance<\/a>\u00a0says. \u201cThe system must present operators with realistic representations of communications availability and autonomous behavior, preventing false assumptions of control.\u201d<\/p>\n DIU announced the Orchestrator project on Jan. 13. The next day, the Pentagon\u2019s Research and Engineering branch, led by\u00a0newly empowered Under Secretary Emil Michael<\/a>,\u00a0released video<\/a>\u00a0from what appeared to be a related effort, the military\u2019s first \u201cSwarm Forge<\/a>\u201d experiment on Jan. 8. While less than a minute and a half, the video and voiceover portrayed a four-drone formation \u2014 one \u201cbrain\u201d drone leading three expendable kamikazes \u2014 striking three simulated tanks in seconds, almost simultaneously.<\/p>\n \u201cThe leader drone will identify what needs to be hit, and then the three follower drones will go in and attack,\u201d an unidentified narrator explained. In this\u00a0specific experiment, defense contractor Auterion\u00a0said<\/a>\u00a0its software coordinated the swarm and told Breaking Defense that a single human operator designated the targets, and then left the details up to the AI.<\/p>\n