Cover story: Jiang J, Zhou X, Chen R, et al. Embodied AI for Aerial Manipulation: A Survey. Technology 1, 260007 (2026).

Here, we review recent advances in aerial manipulation from the perspective of embodied intelligence, where aerial robots evolve from mobile sensing platforms into physical agents capable of actively interacting with the environment. We organize the field along three interconnected dimensions: manipulation tasks, aerial embodiments, and intelligent methodologies. The review examines increasingly complex interaction tasks, representative aerial platforms and manipulation mechanisms, and intelligent approaches ranging from model-driven and human-input-driven methods to data-driven and hybrid model–data-driven paradigms. Particular attention is given to emerging learning-based methods, including reinforcement and imitation learning, vision-language-action models, and world models, which offer new possibilities for semantic reasoning, adaptive interaction, and long-horizon manipulation. We also briefly discuss swarm intelligence for cooperative aerial manipulation, where multi-robot coordination can extend the capability and robustness of embodied aerial systems. Finally, we highlight key challenges in embodiment diversity, aerial-domain data, and sim-to-real transfer, and outline future directions toward more autonomous, robust, and generalizable aerial manipulation for real-world applications.

Explore All Journals

Policies

Export File

Citation

Format

Content

WeChat Qrcode