When　engaging　in　tasks　involving　dynamic　target　tracking　and　physical　interactions,　unmanned　aerial　manipulator　(UAM)　commonly　encounter　challenges,　such　as　the　difficulty　of　obtaining　speed　information,　slow　tracking　control　speeds　and　strong　disturbances　during　physical　interactions.　This　paper　presents　a　series　of　solutions　to　address　these　challenges.　Firstly,　an　image-based　adaptive　velocity　observer　is　introduced　to　mitigate　the　issue　of　acquiring　speed　information.　By　estimating　the　relative　velocity　between　the　UAM　and　dynamic　targets,　this　approach　reduces　noise　and　errors　in　speed　measurements.　Subsequently,　to　address　the　problem　of　slow　tracking　control　speeds,　a　non-singular　fast　terminal　sliding　mode　visual　servo　control　method　is　proposed,　ensuring　stable　tracking　of　dynamic　targets　within　a　finite　time　frame.　Moreover,　considering　the　significant　perturbations　introduced　by　onboard　manipulators　during　physical　contact　operations,　a　backstepping　admittance　control　strategy　with　adaptive　upper-bound　estimation　is　proposed　for　the　onboard　manipulator　to　achieve　compliant　behavior　while　minimizing　the　influence　of　force　sensor　noise.　Finally,　the　effectiveness　of　our　approach　is　validated　through　dynamic　target　tracking　and　physical　interaction　experiments　conducted　in　both　simulated　and　real-world　scenarios.　The　results　demonstrate　the　capability　to　achieve　rapid　visual　servoing　tracking　performance　for　UAMs　without　relying　on　relative　velocity　information　of　dynamic　targets　and　the　ability　to　maintain　stable　control　performance　during　physical　interactions　with　humans.　©　2025　South　China　University　of　Technology.　All　rights　reserved.