Underwater swimming manipulator ( USM), as a novel type of underwater robot, plays a significant role in the smooth completion of underwater observation and operation tasks due to its stable control. This article addresses issues such as the difficulty in accurately establishing the model, the challenge in observing external disturbances like water currents, and the difficulty in tuning the parameters of the controller for all six degrees of freedom in the stabilization control of the USM. The study proposes a control method based on fuzzy feedback linearization and an adaptive extended state observer. Firstly, the dynamic model of the USM is formulated. Secondly, considering the strong nonlinearity and strong coupling characteristics of the USM, a PD controller is designed after feedback linearization of the dynamic model, and fuzzy control logic is introduced to adjust the controller parameters. To reduce the impact of modeling errors and various uncertain disturbances present in the actual underwater environment on the control effect, an adaptive extended state observer is designed to compensate for inaccuracies of the model and external disturbances in real time. Ultimately, during lake trials, the six-dimensional root mean square error of this scheme is 0. 425 2, 0. 166 8, 0. 168 5, 0. 267 4, 0. 117 4 and 1. 003 3, which shows a significant improvement compared to traditional schemes. It shows the effectiveness of the proposed solution.