GreenBotAI - Frugal and adaptive AI for flexible industrial Robotics
Partenaires : Grai Matter Labs (coordinateur), ENSAM, Fraunhofer IWU, Univ. Munchen
Membres LISPEN : Richard Béarée, Adel Olabi, Stéphane Thiery, Adrien Florit, Vincent Deschodt
Le projet GreenBotAI est un des 5 lauréats de l’appel à projet « Franco-German Innovation projects on artificial intelligence technologies for risk prevention, crisis management and resilience » supporté par les ministères de l’économie Français (via BPI) et Allemand (via DLR).
Abstract : GreenBotAI project has three main objectives, ensure continuous production in Europe during pandemics, ensure sovereignty to Europe regarding production automation and to reduce the environmental impact of European factories – by reducing robotic applications energy consumption by 50%. The innovations that need to be developed concern the AI processor and the AI algorithms for AI-based cognitive robot programming. The proposed solution is a combination of AI-based cognitive robot programming, an AI processor optimized for ultra-low latency / low power processing, and off-the-shelf robotics components. The solution will take on the form of automated cells with robots or several robots in a common workspace for handling and/or assembly of different parts, and will address the following functionalities: a) Self-learning of part placement and real-time visual servoing (including incremental learning), b) Workspace monitoring (path planning for collision avoidance of several robots, flexible assembly and Human-robot collaboration), c) Determination of best gripping and holding configuration according to the part and the available end-effector, d) Real-time-based quality inspection of parts and force-torque controlled assembly under changing conditions, e) Consolidated evaluation of collected data in the cloud.
