CFP: 1st Frontier Workshop on Representations for Physical AI
Cross-AI
RePAI 2026: 1st Frontier Workshop on Representations for Physical AI
RePAI 2026 will be held online on Dec. 14-16.
Embodied intelligence emerges through the interaction of an agent, its body, and its environment. Advances in foundation models, generative learning, simulation, robotics, and autonomous systems are enabling increasingly capable Physical AI systems, yet the representations that support physical intelligence remain poorly understood.
The Representation for Physical AI (RePAI) workshop positions representation as a fundamental scientific problem in embodied and physical intelligence. Physical AI systems operate across levels of abstraction, from high-level semantic and task-level reasoning to real-time physical dynamics and low-level motor control. Effective representations must connect what an agent perceives with what it can predict, reason about, and change. This requires representations of objects, agents, semantics, physical properties, causality, actionable affordances, system constraints, and long-horizon consequences.
To address this challenge, RePAI brings together researchers across robotics, machine learning, computer vision, natural language processing, cognitive science, neuroscience, physics, and dynamical systems to study how representations for Physical AI are structured, grounded, learned, transformed, and evaluated for reliable interaction and action.
A central frontier of this workshop is the interface between semantic and physical intelligence: connecting data-driven foundation models with world models, predictive dynamics, planning, and classical control. RePAI also emphasizes representations that transfer across environments and embodiments and remain robust under distribution shift and physical interaction.
RePAI seeks to advance a view of embodied representation in which success is measured not only by prediction or recognition, but by the ability of representations to support generalization, intervention, planning, and reliable physical action.
Important Dates (AoE)
October 26, 2026Full/Short Paper Submission
November 2, 2026 Poster/Demo Paper Submission