Embodied intelligence for crop production

Global agriculture faces growing food demand, labor shortages, and climate uncertainty. Although automation has improved agricultural operations, adapting to variable crop states and unstructured environments remains challenging. Embodied intelligence integrates multimodal perception, contextual reasoning, and adaptive action through continuous interaction with crops and their environment. This review examines advances in embodied perception, decision-making, and execution for crop production. We discuss how multimodal sensing, crop growth models, and adaptive control connect individual operations with management objectives across the crop growth cycle. We further identify priorities for system integration, long-term validation, and practical deployment to support precise, sustainable, and resilient food production.