In order to improve the ecological compatibility and data acquisition efficiency of aquaculture underwater observation， a bionic underwater observation device with low disturbance and high stability was designed. In this experiment， the morphological design of bionic manta rays was adopted， the outline of biological specimens was extracted based on computer vision，the external morphological model of the NACA0013 airfoil was constructed， and the radial kinematic model was established by combining the Lighthill theory. The hydrodynamic performance was verified by the moving tank test， and the actual water area test was carried out to evaluate the degree of disturbance to the fish population and the image transmission ability. The results show that the goodness of fit between the biomimetic body contour and the NACA0013 airfoil is 0.944， which has excellent hydrodynamic characteristics， and the thrust， lift and pitch moment values reach the maximum when θ=50° and f=0.830 Hz in the moving tank test， and the oscillation is regular， indicating the stability of the motion. After testing， the average tracking rate of fish schools is as high as 36.5%， which proves that the device has minimal interference with fish schools and excellent environmental compatibility， and can transmit high-definition images of the integration of bionic manta rays and natural fish in real time， which fully meets the needs of ecological observation. The results show that NACA0013 airfoil has a significant streamlined fitting advantage. The thrust， lift and pitching moments of the bionic manta rays increased with the increase of motion amplitude and frequency， but were significantly affected by the environmental flow velocity. When the movement speed of the bionic manta ray was 0.3 m/s， the fish follow-up rate reached the peak. The device can realize non-disturbance detection in aquaculture and provide effective data support for aquaculture management. This study provides a new eco-friendly underwater observation equipment for aquaculture， and its bionic design and intelligent control methods can be extended to the fields of marine ecological monitoring and underwater robots， so as to promote the development of low-interference underwater technology.