This dissertation is supported by the General Program of National Natural Science Foundation of China, titled “Research on mapping of underwater acoustic channels using multiple underwater gliders” under Grant 61673370. The key issues are to understand the noise characteristics of underwater vehicles, to design the optimal sampling path of underwater vehicles, to construct a spatial field from sampled measurements, and to precisely navigate and localize underwater vehicles for a more accurate reconstruction. Considering that underwater acoustic experiments cannot be frequently carried out, a simulation system based on the ocean-acoustic coupled model is constructed. This dissertation mainly includes the following contents. 1. Constructing a simulation system based on the ocean-acoustic coupled model. The temperature and salinity data from a developed ocean model is input into the BELLHOP 3D model, which realizes end-to-end automatic operation from the simulation area selection to the acoustic parameter output in a simulated ocean environment, supporting the algorithm verification in the subsequent research. 2. Noise characteristics analysis of underwater vehicles. We analyze radiated noise characteristics of a developed “Sea Whale 2000” autonomous underwater vehicle (AUV) with no installed hydrophone to provide a reference for potential acoustic applications. For the developed Sea-Wing acoustic underwater glider, self-noise characteristics, including hydrodynamic ?ow noise and mechanical noise, are analyzed. Based on the acquired knowledge, a joint convolution ?ltering and thresholding method is proposed to remove the glider noise from noisy data recorded during sea trials. All glider noise could be removed from the recorded data. 3. Research on an optimized sampling strategy of underwater vehicles based on compressive sensing (CS). CS algorithm is firstly introduced into the acoustic field reconstruction. Then aiming at the problem that general lawnmower trajectories cannot adequately sample the coherent structure of acoustic fields, a genetic algorithm-based measurement matrix optimization method that combines the restricted isometry property (RIP) of CS and the continuous motion constraints of underwater vehicles is proposed to improve the reconstruction accuracy. 4. Research on a kriged compressive sensing (KCS) approach to reconstruct acoustic fields from measurements collected by underwater vehicles. Aiming at the problem that the continuous sampling characteristics of underwater vehicles along trajectories affects the coherence between the sparse transformation matrix and measurement matrix, a KCS approach is established by adding random virtual samples from kriging estimated fields to make the sensing matrix satisfy the RIP as much as possible. A weighting function is then introduced to distinguish between virtual samples and real samples, then a weighted KCS approach is proposed to further improve the reconstruction accuracy. 5. Research on navigation and localization algorithms for underwater gliders. Acoustic field sampling relies on the spatial position of underwater vehicles. To improve the sample position accuracy, the vehicles’ navigation and positioning performance needs to be improved. An extended Kalman Filter (EKF)-based method, combining a motion model of underwater gliders with acoustic measurements from a single beacon, is proposed to estimate the glider positions in a predict-update cycle. Two parameters, the attack and drift angles, calculated based on the coef?cients of hydrodynamic forces are introduced into the glider model. The travel-time differences between signals received from a single beacon, multiplied by the sound speed, are taken as the measurements. To improve the EKF estimate, the RTS smoothing algorithm is adopted for each gliding cycle by introducing subsequent measurements.