29 mars 2019 | International, C4ISR, Autre défense

DARPA: Using AI to Build Better Human-Machine Teams

The inability of artificial intelligence (AI) to represent and model human partners is the single biggest challenge preventing effective human-machine teaming today. Current AI agents are able to respond to commands and follow through on instructions that are within their training, but are unable to understand intentions, expectations, emotions, and other aspects of social intelligence that are inherent to their human counterparts. This lack of understanding stymies efforts to create safe, efficient, and productive human-machine collaboration.

“As humans, we are able to infer unobservable states, such as situational beliefs and goals, and use those to predict the subsequent actions, reactions, or needs of another individual,” said Dr. Joshua Elliott, a program manager in DARPA's Information Innovation Office (I2O). “Machines need to be able to do the same if we expect them to collaborate with us in a useful and effective way or serve as trusted members of a team.”

Teaching machines social intelligence however is no small feat. Humans intuitively build mental models of the world around them that include approximations of the mental models of other humans – a skill called Theory of Mind (ToM). Humans use their ToM skill to infer the mental states of their teammates from observed actions and context, and are able to predict future actions based on those inferences. These models are built on each individual's existing sets of experiences, observations, and beliefs. Within a team setting, humans build shared mental models by aligning around key aspects of their environment, team, and strategies. ToM and shared mental models are key elements of human social intelligence that work together to enable effective human collaboration.

DARPA's Artificial Social Intelligence for Successful Teams (ASIST) program seeks to develop foundational AI theory and systems that demonstrate the basic machine social skills necessary to facilitate effective machine-human collaboration. ASIST aims to create AI agents that demonstrate a Machine ToM, as well as the ability to participate effectively in a team by observing and understanding their environment and human partners, developing useful context-aware actions, and executing those actions at appropriate times.

The agents developed under ASIST will need to operate across a number of scenarios, environments, and other variable circumstances, making the ability for them to evolve and adapt as needed critical. As such, ASIST will work to develop agents that can operate in increasingly complex environments, adapt to sudden change, and use observations to develop complex inferences and predictions.

During the first phase of the program, ASIST plans to conduct experiments with single human-machine interactions to see how well the agents can infer human goals and situational awareness, using those insights to then predict their teammate's actions and provide useful recommended actions. As the program progresses, the complexity will increase with teams of up to 10 members interacting with the AI agents. During these experiments, ASIST will test the agents' ability to understand the cognitive model of the team – not just that of a single human – and use that understanding to develop appropriate situationally relevant actions.

Full details on the program can be found in the Broad Agency Announcement (BAA) solicitation, which has been posted to the Federal Business Opportunities website, https://www.fbo.gov/index?s=opportunity&mode=form&id=9d4acf0aba98916288a541bd07810004&tab=core&_cview=1

https://www.darpa.mil/news-events/2019-03-21b

Sur le même sujet

  • Germany walks away from Lockheed, Boeing cargo helicopter offers

    30 septembre 2020 | International, Aérospatial

    Germany walks away from Lockheed, Boeing cargo helicopter offers

    Sebastian Sprenger COLOGNE, Germany — The German Defence Ministry wants to start fresh on a new heavy transport helicopter after finding that offers from American firms Boeing and Lockheed Martin for the Chinook and the King Stallion, respectively, were too expensive, officials announced Tuesday. The surprise decision halts an acquisition race that was scheduled to a see a contract awarded in 2021. The two companies delivered their initial proposals for the program, aimed at replacing Germany's aging CH-53G helicopters, in January 2020. A request for a second proposal was expected by the end of this year. The move comes as the budget implications of the coronavirus crisis are starting to materialize, despite leaders' pledges to keep military spending high. Given that context, other programs are also expected to be on shaky ground, according to sources in Berlin. Officials canceled the helicopter race — locally known as Schwerer Transporthubschrauber, or STH — because the government deemed offers by the defense giants as “uneconomical,” a Defence Ministry statement read. The chance of meeting all requirements while adhering to the envisioned multibillion-dollar budget would be “unlikely,” officials wrote. The STH program was initially planned to be a poster child for a no-frills, off-the-shelf purchase that's easy on the defense budget. But acquisition officials kept piling on requirements to such a degree that it surprised some industry executives associated with the bidders, Defense News reported earlier this year. Also included in the government's requirements was a decades-long maintenance scheme under which contractors had to guarantee certain availability rates. Tobias Lindner, the Green Party's point man in the Bundestag for defense issues, called the helicopter program's a “bitter sign” for Germany's soldiers. He argued the Defence Ministry had been naive in its approach to the much-needed acquisition. “A new competition alone won't solve that problem,” he said. https://www.defensenews.com/global/europe/2020/09/29/germany-walks-away-from-lockheed-boeing-cargo-helicopter-offers/

  • Hypersonic threats need an offense-defense mix

    5 août 2019 | International, Aérospatial

    Hypersonic threats need an offense-defense mix

    By: Melanie Marlowe Next week, people from across the missile defense community will gather at an annual symposium in Huntsville, Alabama, to consider how to adapt U.S. missile defense efforts to the challenge of renewed competition with Russia and China. A centerpiece of their discussions will be the emergence of advanced hypersonic missile threats and what to do about them. Over the past few years, the Pentagon has prioritized the development of offensive hypersonic strike weapons, with billions of dollars in contracts already awarded for each of the major military services to acquire hypersonic strike missiles of their own. The counter-hypersonic mission, however, received surprisingly short shrift in recent defense budgets, with progress on hypersonic defense thus far piecemeal and halting. Some leading military officials charged with procuring hypersonic strike missiles have said that defending against hypersonic missiles is too hard, so we shouldn't even try. That short-sighted approach is at odds with the vision of newly confirmed Secretary of Defense Mark Esper, who stated to Congress that he will advocate hypersonic missile defense, to include the development of new sensors, interceptors, and advanced command-and-control systems. Public commentary on hypersonic threats has been somewhat hyperbolic. Yes, hypersonics are fast — five or more times the speed of sound — but that's slower than many ballistic missiles. Aerodynamic maneuver makes for a less predictable flight path, but this also means that atmospheric friction would remove the kind of decoys that might accompany a ballistic reentry vehicle. Whether a boosted glide vehicle, a scramjet cruise missile or a maneuverable quasi-ballistic missile, hypersonics pose a complex air defense challenge, but they are not invulnerable. The strategic significance of hypersonics is nevertheless quite real. Today's Patriot, Terminal High Altitude Area Defense and Aegis defenses protect American carrier groups and ground forces against aerial and ballistic missile attack. Designed to go around or under those defenses, hypersonics are a more sophisticated means to hold forces at risk, and thereby undermine our broader defense goals and alliance system. Even if the United States catches up with the Chinese and Russians on hypersonic strike, our adversaries' ability to hold U.S. carriers and forward bases at risk will push back U.S. forces. They could certainly also be used to target the American homeland, but the more urgent threat is regional. Passive defense only goes so far — ships can only go so fast, and air bases cannot be moved. Active defenses must be part of a balanced strategy. The first priority here is a space sensor layer. Unlike ballistic missiles, hypersonic missiles fly at lower and changing altitudes, are harder to see, and travel an uncertain flight path. Current early warning satellites can detect the launch of boost-glide vehicles but are unsuited to tracking them during the glide phase. Today's surface-based ballistic missile radars would only be able to spot a weapon once it crosses the horizon. Only space sensors can provide birth-to-death, fire-control quality tracks for hypersonic missiles. Unfortunately, recent budget requests have been rather tepid in their commitment to space sensors. The administration's 2020 request virtually divested the program, and for the second year in a row the Missile Defense Agency listed the space sensor layer as its No. 1 unfunded priority. Thankfully, Congress seems to be in the process of restoring $108 million to return the program to the MDA to move out on development this year. The second element of hypersonic defense is interceptors. Although existing interceptors may well be improved, Secretary Esper has affirmed that new interceptors will have to be developed that are better suited to the mission's stressing thermal and high-maneuver environment. The MDA's third-highest unfunded priority for 2020 — $720 million for hypersonic defense — seems unlikely to be restored this year, but should be restored in the 2021 budget. Directed-energy weapons could potentially target hypersonic threats in their cruise phase or jam them in their terminal phase, but the mission's complexity will almost certainly require both kinetic and nonkinetic effectors. The most challenging element will be developing a command-and-control architecture that ties everything together. A long-range hypersonic glide vehicle of significant range could cross continents and multiple combatant commands. Even with better interceptors and an adequate sensor layer, information and fire-control solutions must be developed and rapidly passed to commanders. The Command and Control, Battle Management, and Communications network that supports the Ballistic Missile Defense System may be the foundation of such an architecture, but more dramatic upgrades will be required. The advent of the hypersonic era is central to the efforts by Russia and China to counter U.S. power projection in the world. The Pentagon's recent focus on hypersonic strike is necessary but insufficient. It falls now to congressional leadership and those assembling the 2021 budget to rebalance it with a more appropriate mix of hypersonic strike and defense. https://www.defensenews.com/opinion/commentary/2019/08/02/hypersonic-threats-need-an-offense-defense-mix/

  • Defense bill underlines need for tactically responsive space program

    9 décembre 2022 | International, Aérospatial

    Defense bill underlines need for tactically responsive space program

    The language continues a three-year push from lawmakers for the service to prioritize the capability.

Toutes les nouvelles