PhD Scholarship Programme Vicente López: Researcher in Generative Multimodal World Modeling and[...]

Overview PhD Scholarship Programme Vicente López: Researcher in Generative Multimodal World Modeling and Predictive Reasoning for Autonomous Robots at Eurecat - Technology Centre. Join to apply for the PhD Scholarship Programme Vicente López: Researcher in Generative Multimodal World Modeling and Predictive Reasoning for Autonomous Robots role at Eurecat - Technology Centre. About Eurecat Robotics and Automation Unit Eurecat Robotics and Automation Unit specializes in pioneering research aimed at enabling intelligent robotic systems that can dynamically adapt and reliably operate within complex human-centric environments. With a multidisciplinary team of experts in artificial intelligence, robotics, and human-robot interaction, the unit integrates cutting-edge generative AI models, advanced cognitive robotics methodologies, and state-of-the-art multimodal perception technologies. Eurecat’s research outcomes span applications including assistive technologies, healthcare robotics, industrial automation, and urban service robots. The team combines deep learning, symbolic reasoning, reinforcement learning, and neuro-symbolic hybrid methods to ensure robust, adaptable, and scalable solutions for real-world deployment. Eurecat Robotics and Automation Unit, in collaboration with Universitat Politècnica de Catalunya, has opened a position for a PhD candidate to develop a thesis in the field of Generative Multimodal World Modeling and Predictive Reasoning for Autonomous Robots. This thesis aims to pioneer a comprehensive robotic world-modeling framework leveraging state-of-the-art generative AI techniques (Vision-Language Models, Diffusion Models) and advanced multimodal methods. The core objective is to establish predictive models that robustly represent both physical and semantic dimensions of the environment. By synthesizing multimodal sensory inputs (visual, linguistic, tactile, and spatial), the research will enable robots to accurately predict outcomes of their actions, ground generative model outputs in real-world sensor data, and dynamically reason about potential future states. Emphasis will be placed on developing innovative knowledge representations using graph-based neural networks to facilitate real-time updates and intelligent decision-making. The outcome will significantly enhance robotic autonomy and adaptability in complex, uncertain, and human-centric scenarios. Functions and Responsibilities Literature Review & State-of-the-Art Analysis: Comprehensive analysis of generative modeling (VLMs, diffusion) and multimodal prediction methods. Generative Multimodal World Modeling: Build generative world models that integrate visual, linguistic, tactile, and forces data for richer, context-aware environmental representations. Predictive Consequence Reasoning: Integrate multimodal world models with reasoning modules to predict and evaluate the consequences of actions to support planning in open-world settings. Semantic and Physical Grounding: Design hybrid grounding methods combining generative VLM outputs with real-time sensor data for accurate understanding and actionable interpretation. Modern Knowledge Representation: Develop dynamic knowledge representations using Graph Neural Networks (GNNs) and Graph Transformers for real-time updates and efficient retrieval. System Integration & Validation: Integrate algorithms into robotic platforms and validate performance in simulated and real-world open-domain scenarios (e.g., domestic, healthcare, urban environments). Publications & Collaborations: Publish findings in top-tier robotics/AI conferences (ICRA, RSS, CoRL) and collaborate with research partners in the Catalan & European research ecosystem. Offer Full-time contract Duration: 3 years 27 days of paid holiday leave Flexible schedule Reduced work hours on Fridays and during the summer Access to a wellness platform Annual gross compensation and coverage of university PhD enrollment fees Flexible benefits package PhD training program Requirements Studies and Knowledge MSc degree in Robotics, AI, Machine Learning, or related fields. Strong background in deep learning, generative models (VLMs, diffusion models), multimodal AI, and robotics. Experience Proficient in Python, PyTorch/TensorFlow. Experience in robotics frameworks (ROS/ROS2) and familiarity with multimodal data processing (vision, language, tactile). Languages Excellent oral and written communication skills in English. Knowledge of Catalan and/or Spanish is considered an asset. Submission of Applications To formalize their application, applicants must: Complete the registration forms with all requested information and provide the following documents: photocopy of valid NIF/NIE or passport, CV and cover letter, photocopy of the official academic certificate or