Description du projet
Une solution SLAM pour des utilisations en environnement réel
Des smartphones à la réalité augmentée (RA), la localisation et la cartographie simultanées (SLAM) promettent une révolution. Ce système permet de déterminer la position et l’orientation de notre appareil afin de créer une carte de l’environnement alimentée par une entrée caméra. Le projet SLAM4AR, financé par l’UE, développera de nouveaux algorithmes de mouvements d’une caméra pour permettre une reconstitution 3D précise de l’environnement observé. Cette nouvelle technologie permettra en outre l’insertion d’objets dans l’environnement 3D. Ses applications potentielles en environnement réel sont nombreuses et vont de l’éducation (montrer aux étudiants en médecine l’emplacement en 3D des os et des organes internes) à l’industrie (les mécaniciens automobiles pourront visualiser les structures internes d’un moteur) et à la culture (aider les touristes à se rendre à un endroit précis d’un musée).
Objectif
The PoC Project SLAM4AR is focused on developing technological foundations for advanced augmented reality (AR) applications on smartphones. According to a recent study of Hampleton & Partners, the market value for AR applications is expected to break the $170 billion barrier by 2022. And a significant share of this market value will come from smartphone applications. In the PoC project SLAM4AR, we will build up on state-of-the-art algorithms for visual Simultaneous Localization and Mapping (SLAM) that we developed in the ERC CoG “3D Reloaded”. These algorithms allow one to recover the motion of a camera and a 3D reconstruction of the observed environment at unprecedented precision, robustness and large-scale capability (among existing real-time capable algorithms). We will reformulate and streamline these algorithms so that they can run on smartphones. They will leverage the smartphone’s camera and inertial sensor in order to compute in real-time both the location of the phone and a 3D map of the environment. In several AR applications, we will demonstrate that our algorithms with their superior precision and robustness serve as a key enabler for advanced AR technology. For example, we can perform object insertion (beyond Pokemon Go) in a way that dynamic objects interact more naturally and faithfully with a complex 3D environment – they accurately sit on chairs or tables, or they convincingly roll down slopes, etc. This will enable numerous AR applications such as teaching medical students about the 3D location of bones and inner organs, training novice car mechanics in the inner structures of a motor. Or navigating tourists to a desired location in a museum or a church. There is a rapidly growing market for AR applications on phones and we are convinced that our smartphone-based visual SLAM technology will serve as a key enabler.
Champ scientifique
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsensorsoptical sensors
- engineering and technologyelectrical engineering, electronic engineering, information engineeringinformation engineeringtelecommunicationsmobile phones
- natural sciencescomputer and information sciencessoftwaresoftware applicationssimulation software
Programme(s)
Régime de financement
ERC-POC-LS - ERC Proof of Concept Lump Sum PilotInstitution d’accueil
80333 Muenchen
Allemagne