Embodied Cross-Domain Intelligence in Biomedical Microrobots: A Review

Microrobots are emerging as transformative tools for biomedical applications, including minimally invasive diagnostics, targeted therapy, and microsurgical intervention. However, achieving reliable performance and adaptation to diverse tasks and environments requires capabilities that exceed any single form of intelligence. Here, we introduce embodied cross-domain intelligence, a framework for synergistic coupling across physical, biological, computational, and human intelligence, enabling multifunctional, collaborative, and adaptive microrobotic behavior in dynamic, safety-critical biological settings. Unlike prior reviews that address these intelligence domains in isolation, this review aims to provide a unified framework, outlining each domain's principles, recent advances, and limitations; analyzing the interfaces that foster synergy; and mapping representative domain combinations to major biomedical applications. We further identify core challenges in integration, control, safety, and validation and outline future research directions to accelerate clinical translation. By framing biomedical microrobot development as a cross-domain synergy challenge, this review aims to guide interdisciplinary efforts toward systems capable of executing complex, multistage tasks across their operational life cycle. The associated project is available on the online project page (https://nuounuou.github.io/Embodied-Cross-Domain-Intelligence-in-Biomedical-Microrobots-A-Review/).