Le projet Autonomie des Mobilités Aériennes du Futur a pour objectif de structurer l’autonomie des nouvelles mobilités aériennes (Regional Air Mobility / Urban Air Mobility) au Québec à travers deux approches complémentaires. D’abord, en capitalisant sur le riche écosystème technologique pour faire du Québec la place forte des systèmes d’aviation numériques.

Virtual and mixed reality (VR/MR) systems have burgeoned over the last couple of years with applications in healthcare, gaming, telepresence, and skills training, to name a few. Within the skills training sector, for example, police/law enforcement training is an important application domain which has seen increased adoption worldwide. In Canada, Public Safety and the Royal Canadian Mounted Police (RCMP) have invested millions of dollars to set up state-of-the-art VR/MR facilities to train the next generation of law enforcement agents.

Les progrès réalisés au cours des dernières années dans le domaine des véhicules intelligents ont fait des bonds de géant en termes d'automatisation et de connectivité. Ces avancées ouvrent les véhicules intelligents à des cyber-attaques potentielles qui peuvent mettre en danger les informations personnelles et les fonctions essentielles du véhicule. Conscient de ces défis, Thales Digital Solutions Inc.

Thales Canada is developing Refit Optimizer, a prototype solution for managing military ship refit projects and maintenance operations. In particular, this tool offers a scheduling optimization component, which decides when each task needs to be executed. During its development, three different criteria were identified according to the challenges faced by the shipyards: minimization of the project's duration, minimization of the costs associated with overtime labor, maximization of the schedule's robustness.

Depuis maintenant 3 ans, Thales a lancé son accélérateur AI@Centech. Situé à Montréal, l’accélérateur a pour objectif d’accompagner, sous forme de cohortes de 6 mois, des start-ups qui fournissent des solutions basées sur l’intelligence artificielle (AI). Cette grappe Mitacs, qui réunit des stagiaires de l’ÉTS et HEC, vient appuyer Thales dans la structuration et l’expansion de son accélérateur d’un point de vue de gestion de l’innovation.

Engineering organisations like Thales rely on large quantities of technical knowledge. When resolving a technical problem, for example, users have to follow a multi-step procedure in which the steps are described with various levels of detail, may not be up to date, or may not target the exact problem they are facing. In this context, our project aims at developing AI language and knowledge representation models that will represent technical knowledge from an unstructured dataset of procedures, in order to identify missing knowledge through a question-answering process.

Reinforcement learning algorithms are successfully employed in diverse industries, for instance in autonomous driving, trading or gaming. However, their generalization especially in critical-decision making systems has raised concerns about their robustness to attacks. The 22nd recommendation issued by the White House (2016) to prepare for the future of AI states: ”Agencies (...) should ensure that AI systems and ecosystems are secure and resilient to intelligent opponents”.

Thales développe le Hospital Optimizer, un outil pour gérer l’utilisation des salles d’opérations en les affectant à différentes équipes médicales. Cette allocation tient compte de la demande, des disponibilités médecins, de la disponibilité des salles et des types de chirurgies pouvant être effectuées dans chaque salle. Si la version courante du Hospital Optimizer prend bien en compte ces éléments du problème (et même plus), il lui manque toutefois une fonctionnalité cruciale afin de permettre l’acceptabilité de cet outil dans le milieu de travail : la production d’horaires récurrents.

Children and animals learn by trials and errors from their environments. They don’t need many similar examples to learn how to do a task since they develop self-awareness about a given experience in order to avoid repeated regrets caused by failed trails. A robot machine can be programmed to develop such awareness by using reinforcement learning (RL). However, the RL algorithms require collecting a large amount of data to reach high performance, which makes them only successful in simulation environments.

This project aims at proposing a novel approach for the task of Conversational and Visual Question Answering. We propose to use Knowledge Bases for questions which require common sense, or basic factual knowledge from external structured content. We will explore the application, adaptability and utility of common sense knowledge to handle the question answering problem on both textual and visual questions.