The ocean area occupies a large part of the earth’s area, how to use underwater robots to carry out deep-sea exploration tasks has become an urgent problem in the field of marine resources. In this paper, we design a small AUV underwater robot from five aspects: power supply, control motherboard, power, communication, and sensor. Due to the complexity of the underwater environment during deep-sea exploration, the sensors of the underwater robot need to be calibrated to facilitate data collection and acquisition. To acquire and process underwater image data, the robot uses optical and acoustic imaging principles. The bitmap algorithm is employed to construct a multi-sensor fusion model for depth detection, which is then analyzed for application. The underwater robot is basically able to reach the specified position quickly and smoothly according to the set motion, and its underwater robot infiltrating the bitmap algorithm (x-axis time of 89s, y-axis time of 98s, and z-axis time of 99s) is obviously faster than that of the traditional SLAM algorithm in terms of convergence speed. The underwater robot navigates approximately 400 m in advance of the deep-sea probe’s localization, which allows for fast and stable error convergence. This study meets the covertness requirements of underwater robots when performing detection tasks, and can achieve independent and autonomous navigation of underwater robots, which has a good application prospect.