Particle filter ROV navigation using hydroacoustic position and speed log measurements

An integrated navigation system design is presented for an underwater remotely operated vehicle (ROV). The available navigation information is an acoustic position measurement and a Doppler log speed measurement. Both measurements are studied in detail and modeled statistically. A kinematic model is assigned to the ROV with its driving noise from a Gaussian mixture, and a particle filter is suggested to estimate ROV position and velocity. The advantages of using a particle filter in this ROV navigation scheme are: 1) to make full use of all available information to improve the estimation performance, such as the speed measurement that is a nonlinear function of the states; 2) the particle filter makes good use of a Gaussian mixture as the driving noise, which makes the ROV kinematic model more realistic in both high and low frequency ranges; 3) a good estimate of the ROV velocity vector is achieved. The algorithm of the particle filter is presented and verified through a simulation based on real data. This shows that the estimation performance of the particle filter is clearly better than that of a Kalman filter.