This comprehensive review explores Embodied Multimodal Large Models (EMLMs) that integrate perception, cognition, and action capabilities to navigate and interact with real-world environments.

• Examines the evolution of EMLMs, building on Large Language Models (LLMs) and Large Vision Models (LVMs)
• Addresses key challenges in embodied perception, navigation, and decision-making
• Analyzes datasets and benchmarks essential for training robust embodied AI systems
• Identifies promising future research directions for engineering more capable autonomous systems

For engineering teams, this research provides critical insights into developing AI systems that can perceive their surroundings, reason about complex environments, and take appropriate actions—essential capabilities for next-generation autonomous robots and interactive systems.

Exploring Embodied Multimodal Large Models: Development, Datasets, and Future Directions