Back to news

April 3, 2020 | International, Aerospace

A la recherche de l'IA hybride

Vice-président recherche, technologie, innovation du groupe Thales, David Sadek intervenait jeudi 24 octobre à Toulouse au Forum Innovation IA, un événement organisé par Sciences et Avenir pour le groupe Challenges.

Explicabilité, validité, intégrité... " Pour une IA de confiance " : c'était l'intitulé de l'intervention de David Sadek, le vice-président recherche, technologie, innovation du groupe Thales. "Nous aimerions tous pouvoir toujours faire confiance à tous nos outils technologiques. Mais en réalité tout dépend du contexte. Dans le cas des outils reposant sur l'IA, si Netflix me recommande un film que je trouve détestable, ce n'est pas bien grave, on s'en remettra. Mais dans le cas des systèmes critiques tels que l'on en conçoit régulièrement chez Thales pour des applications défense, espace et aéronautique, la question de la confiance est primordiale".

David Sadek énonce trois impératifs, qui sont autant d'axes stratégiques de la feuille de route IA chez Thales, pour aller vers cette " IA de confiance ". Pour qu'une application de l'IA soi digne de confiance, il faut d'abord qu'elle soit capable d'expliquer. Si elle se contente de proposer ou décider sans être capable de dire pourquoi et comment elle est arrivée à cette proposition ou décision, la défiance s'installera. Elle doit donc être capable de répondre à la question : " Pourquoi ? ".

"Si par exemple un copilote IA recommande au pilote humain de virer à 45°, et que ce dernier demande " pourquoi ", le copilote doit pouvoir répondre par exemple : " parce qu'il y a un problème météo (ou une menace ...) droit devant ". Il ne s'agit pas de simplement tracer la décision, de pouvoir indiquer à ses concepteurs quels neurones, quelles couches de neurones, ont fait pencher la balance dans un sens ou un autre, mais bien d'expliquer dans une langue compréhensible immédiatement par le pilote". Ce qui indique qu'à l'aspect intelligence artificielle s'ajoute celui de " l'interaction homme-machine ".

IA connexionniste et IA symbolique

Cela implique, en restant sur cet exemple, une capacité de compréhension et de génération du langage naturel. Mieux, le pilote appréciera de pouvoir dialoguer dans un langage mixte, reposant sur l'oral et le geste (pour désigner un point sur une carte, etc.). Deuxième impératif selon David Sadek : il faut pouvoir démontrer la validité d'une application de l'IA. C'est-à-dire sa conformité aux spécifications. "Le système développé doit faire tout ce l'on attend de lui et rien que ce que l'on attend de lui. Ce qui suppose que l'on sache spécifier très proprement". On connait l'exemple d'un jeu de bataille navale, un " serious game " destiné à former des officiers de la marine, qui gagnait à tous les coups, parce qu'il sacrifiait systématiquement chaque vaisseau touché, pour ne pas ralentir la flotte.

Aucune règle ne prévoyait, n'interdisait ce cas de figure, passablement inacceptable. Troisième impératif pour une " IA de confiance ", selon David Sadek, la responsabilité des systèmes reposant sur l'IA. C'est-à-dire leur conformité aux cadres légaux, réglementaires et moraux. On connait le tendon d'Achille des réseaux de neurones : ce sont des boîtes noires, on voit ce qui entre et ce qui sort, mais on ne sait pas (trop) ce qu'il se passe à l'intérieur. C'est pourquoi David Sadek insiste sur le fait qu'il faut s'intéresser aussi à " l'autre IA ", l'IA symbolique, celle qui repose sur des règles et des raisonnements et qui s'oppose à l'IA connexionniste des réseaux de neurones. Cette IA symbolique, qui fait moins les gros titres, qui n'a pas connu les progrès fulgurants de l'IA des réseaux de neurones, est plus à même de répondre à ces trois exigences : expliquer, valider, responsabilité. C'est pourquoi la réponse à cette triple exigence passe sans doute, entre autres, par " l'IA hybride ", une IA tirant parti de ces deux branches de l'intelligence artificielle.

Par Pierre Vandeginste

https://www.sciencesetavenir.fr/high-tech/intelligence-artificielle/ia-pvdg_138526

On the same subject

  • Naval Group: Talks with Australia still underway for sub buy

    October 3, 2018 | International, Naval

    Naval Group: Talks with Australia still underway for sub buy

    By: Pierre Tran PARIS — Naval Group, a French shipbuilder, sought to play down media reports of a potential collapse of talks for a major contract with Australia for 12 ocean-going submarines, with elections looming next year in the Commonwealth nation. “We are in the process of pursuing the strategic partnering agreement,” a Naval Group spokesman told Defense News. That agreement covers some 30 years and is an extremely close commitment, he added. “It is plain logic that that takes time,” he said. “We are not particularly worried. We remain confident." Naval Group is working under a 2016 design and mobilization contract on the Australian future submarine program, while negotiating the strategic partnering agreement. The latter is a critical contract which moves the French company closer to building the diesel-electric boats in Adelaide under an Australian $50 billion (U.S. $36 billion) deal. There is concern that failure to reach consensus on intellectual property and warranty conditions have held up a signing of the strategic partnering agreement, with time running out, Australian Broadcasting Corporation reported. “Negotiations on that document have stalled and it is feared they may not be resolved before next year's federal election,” ABC reported. Signing the strategic partnering agreement by the end of this year is needed to allow detailed design contracts to be finalized and building of the boats, the broadcaster reported. In Australia, Defense Minister Christopher Pyne sought to calm the waters with an Oct. 1 reply on social media. “There is no delay to the delivery of 12 Future Submarines. Negotiation of strategic partnering agreement is continuing & we will ensure we end up with an equitable & enduring agreement to deliver the capability our servicemen & women need, and get the best deal for Australian taxpayer.” Nigel Pittaway in Melbourne, Australia contributed to this report. https://www.defensenews.com/digital-show-dailies/euronaval/2018/10/02/naval-group-talks-with-australia-still-underway-for-sub-buy

  • U.S. Air Force Plots Fleet Insertion Path For ‘Loyal Wingman’

    March 9, 2020 | International, Aerospace

    U.S. Air Force Plots Fleet Insertion Path For ‘Loyal Wingman’

    The format of the U.S. Air Force's “fireside chat” series is well-understood. A technology pioneer such as Jeff Bezos, Richard Branson or Mark Cuban appears onstage at an Air Force-affiliated event, counsels an audience of pilots and airmen about innovation and, not least, tries not to offend anyone. Elon Musk arrived at the Air Warfare Symposium on Feb. 28 with a different plan. The founder of SpaceX and Tesla, who seems to delight in publicly tweaking established competitors in the space market such as Boeing and Lockheed Martin, sat on the Air Force Association's (AFA) stage and declared that the fighter aircraft—for decades the heart of the Air Force's tactical combat capability—is already irrelevant. “The fighter-jet era has passed,” Musk said, provoking audible gasps and murmurs in an audience peppered with officers clad in flight suits. Lt. Gen. John Thompson, Musk's interviewer, quickly changed the subject. Hours later, Musk clarified in a tweeted reply to Aviation Week that he meant the fighter aircraft remains relevant, just not the pilot onboard. “The competitor [to a manned fighter] should be a drone fighter plane that is remote-controlled by a human, but with its maneuvers augmented by autonomy,” Musk writes. Musk's comments on airpower should be taken with a grain of salt. Although his companies have sought to disrupt the space, automotive and mining industries, Musk has no track record in the aircraft sector. One of his symposium hosts, David Deptula, a retired lieutenant general who is now dean of the AFA's Mitchell Institute, also pointed out in a rapidly published rebuttal in Forbes that Musk's predictions about autonomy are often wrong, even when it concerns the self-driving capabilities of Tesla cars. But Musk's remarks only differed with those of senior Air Force officials at the same event in the details of timing and scope. For over a year, Will Roper, assistant secretary of the Air Force for acquisition, technology and logistics, has championed a vision of future airpower populated by numerous, small batches of autonomous aircraft augmenting manned fighters with specialized capabilities. For the first time, Gen. James Holmes, head of Air Combat Command (ACC), offered a path to introducing such aircraft into the fleet around 2025-27. In the near term, the Air Force is focused on replacing aging F-15C/Ds with a mix of Boeing F-15EXs and Lockheed Martin F-35As. The Air Force decided to add the F-15EX to its inventory last year even as the Air Force Research Laboratory (AFRL) began experimenting with a new class of low-cost aircraft with an “attritable” value. The first such experimental aircraft, the Kratos XQ-58A Valkyrie, in March 2019 completed the first of four flights made to date. Next year, the Air Force plans to fly the XQ-58A or a similar aircraft with an artificial intelligence “brain,” which allows the so-called Skyborg aircraft to learn maneuvers as it flies. Such capabilities are not far from Musk's vision of future air combat, but they are too immature to replace a fleet of F-15Cs on the verge of being grounded; hence, the decision to buy the F-15EX instead. The next opportunity to introduce a new kind of aircraft comes in about 5-8 years, Holmes says. That timing dovetails, perhaps intentionally, with the schedule for maturing aircraft such as the XQ-58A and Skyborg. The Air Force will need to replace hundreds of F-16 Block 25s and Block 30s, which entered production in the mid-1980s. “There's an opportunity there if we want to cut in something new, a low-cost attritable, loyal wingman and the different things that we're looking at and experimenting with,” Holmes says. In late February, Holmes and Roper met to discuss the meaning of a “fighter aircraft” in the future with the Next-Generation Air Dominance (NGAD) program in the backdrop. The program office for NGAD began operations in October, with a focus on inventing a new production process capable of affordably producing small batches of advanced aircraft every 3-5 years. But Air Force officials are still grappling with the definition of basic requirements such as range and payload, as operations in the vast Pacific Ocean dominate the calculations. “The equation and the kind of math that we use for a fighter still works pretty well in the European environment—the range and payload and distance,” Holmes says. “It's not as effective a solution in the Pacific, because of the great distances. So as you look at NGAD and you look at the following programs, I wouldn't expect it to produce things that necessarily look like a traditional fighter.” The exhibit hall at the Air Warfare Symposium offered some clues. Besides the usual displays and posters of F-35s and F-15s, some new concepts appeared. General Atomics Aeronautical Systems Inc. (GA-ASI) showed a concept design called “Defender,” an apparent variant of the Predator C Avenger, armed with air-to-air missiles and infrared search-and-track sensors. The Defender would protect an “outside force” of enablers, such as tankers and surveillance aircraft, from aerial attack while an inside force of stealth bombers and fighters engaged targets downrange, a GA-ASI spokesman says. Kratos, meanwhile, continues working on the XQ-58. The AFRL initially funded five test flights, but despite a crash landing on the third flight, all test objectives were met after the third test, says Steve Fendley, president of Kratos' Unmanned Systems Division. The AFRL now is accelerating the “missionization” of the XQ-58, Fendley says, adding payloads and potentially weapons. The first payload integration will be demonstrated in April, when the XQ-58 serves as a communication conduit between the F-35 and the F-22. Meanwhile, Kratos has started production, with 12 XQ-58s scheduled off the assembly line by the first quarter of 2021. The fleet will be assigned to multiple demonstration programs, funded by several agencies, Fendley says. The XQ-58's performance helps define the new class of aircraft, called “loyal wingman” in the U.S. and “remote carriers” in Europe. A critical feature shared by the XQ-58 and similar aircraft such as the Boeing Airpower Teaming System (ATS) is range. Both are capable of flying 3,000 nm unrefueled, almost three times the range of the F-35. Unlike the ATS, the XQ-58 does not need a runway to land, and instead deploys a parachute. Both aircraft seem unrecognizable from the typical next-generation fighter favored by ACC, but the command is changing its approach, Holmes says. “In the past at Air Combat Command, we would have built something that we call a fighter road map . . . to figure out what our fighter force will look like for the next 30 years,” Holmes said. “What I would rather build is a capabilities road map that shows how we're going to accomplish the missions for the Air Force that we traditionally had done with fighters.” At the same time, Holmes' counterparts in the Air Force Materiel Command (AFMC) are also changing their approach to fighter acquisition. Last October, the AFMC established the Advanced Aircraft Program Executive Office, which is tasked with reinventing the acquisition process for the next class of fighters. A modern fighter is typically developed over a decade and then sustained for several more. For the next generation, the Air Force now prefers to produce multiple aircraft in small batches, in development cycles lasting only five years. The sustainment phase would be minimal, as the aircraft would be phased out after a brief operational career. The approach requires that the Air Force make the design phase profitable for contractors, which now lose money in design and earn profits during the sustainment phase. The approach means paying higher prices up front for the design, but theoretically less overall during the shorter lifespan of the aircraft. The Air Force is still trying to craft the contractual mechanism for such an acquisition approach, says Gen. Arnold Bunch, the head of AFMC. “Industry is going to have to rethink how they want to go do this. They're gonna have to talk to their boards in a different way,” Bunch says. “[Something] we also have to factor into that is: How do I do my cost estimates? How do I do my financial planning? How do I interact with Congress?” https://aviationweek.com/defense-space/aircraft-propulsion/us-air-force-plots-fleet-insertion-path-loyal-wingman

  • Lithuania enlists EMSA’s RPAS services to monitor ship emissions

    April 8, 2021 | International, Aerospace, Naval

    Lithuania enlists EMSA’s RPAS services to monitor ship emissions

    In response to a request from the Environmental Protection Department of Lithuania's Ministry of the Environment, EMSA is providing the country with Remotely Piloted Aircraft System (RPAS) services to assist in monitoring ship emissions, protect the marine environment and improve maritime safety. The services which began on 23 March will run for three months and will see EMSA RPAS being used to calculate the sulphur content of the fuel being used by the passing ships. Sensors on board the RPAS will measure the emissions from the exhaust plumes of vessels travelling in the main shipping lanes and when at anchorage around the Lithuanian port of Klaipeda. The sulphur content of marine fuel in this Sulphur Emission Control Area (SECA) should be no greater than 0.1%. While great interest has been shown in the emissions monitoring capability of EMSA's RPAS service, the Environmental Protection Department will also be working in collaboration with other Lithuanian authorities - including the Navy, MRCC and fisheries control services – to conduct other types of maritime surveillance missions. The aircraft being used for this service is the Schiebel CAMCOPTER® S100 and it is under contract to EMSA via a consortium led by Nordic Unmanned AS. The model is a vertical take-off and landing drone and is fitted with gas sensors and cameras covering optical and infrared spectral ranges to better detect vessel plumes and conduct maritime surveillance as required. All the information is transmitted in real-time to trained users through EMSA's RPAS Data Centre. Records of the emission measurements are encoded automatically into the THETIS-EU information sharing system. This system is operated by EMSA to assist in the enforcement of the EU sulphur directive as well as to support port inspectors when targeting vessels to be inspected. ABOUT RPAS RPAS services, offered free to all EU Member States by EMSA, have been developed to assist in ship emission monitoring and maritime surveillance operations and can operate in all seas surrounding the European Union. RPAS services can provide support to traditional coast guard functions, including search and rescue and pollution prevention and response. The services are offered to Member States individually and as part of EMSA's regional RPAS strategy, which allows multiple coast guard functions in several EU Member States to be supported by one or more RPAS services. Further expansion of RPAS regionally is planned in 2021 and 2022. ABOUT EMSA The European Maritime Safety Agency (EMSA) is a decentralised agency of the EU, based in Lisbon, Portugal. EMSA serves the EU's maritime interests for a safe, secure, green and competitive maritime sector through support for pollution prevention and response, maritime surveillance, safety and security, digitalisation and the provision of integrated maritime services, and technical assistance. Any requests for further information can be sent to: information@emsa.europa.eu

All news