Abstract: Abstract: Aerial manipulations expand from traditional information acquisition platforms into complex interactive systems with active operational capabilities by establishing continuous or transient physical contact with the environment. This evolution from "passive perception" to "active operation" demonstrates broad application prospects in complex interaction scenarios involving large-scale energy and transportation infrastructure, such as bridges, power grids, and oil and gas pipelines. Compared to non-contact flight missions, contact-based operations involve significant dynamic coupling, contact force constraints, and multi-modal motion switching. The resulting high-dimensional nonlinear constraints restrict the feasible solution space for motion planning, thereby constraining system autonomy and stability. To address these challenges, this paper provides a systematic review of research progress in the field of motion planning for aerial manipulations. First, starting from system architecture, we analyze the impact of various flight platform configurations and operational mechanisms (including tethered, rigid-linked, serial, and parallel structures) on the plannable space and operational capability. Second, we summarize system modeling methods oriented toward motion planning, including contact dynamics modeling and various task planning constraints. On this basis, centered on the planning requirements of contact-based tasks, three mainstream motion planning algorithms—sampling-based, optimization-based, and learning-based—are respectively introduced. Their development trends are analyzed, and their applicability and limitations in high-dimensional state search, dynamic constraint handling, and real-time replanning are compared. Finally, the challenges facing aerial manipulations are summarized, and future development trends are discussed.

Key words: Aerial manipulators, Multirotor aerial robots, Motion planning, Physical interaction, System modeling