An Analytical Framework for Predicting the Performance of Autonomous Underwater Vehicle Positioning

In this paper we present a set of simple models to predict the performance of underwater positioning solutions for autonomous underwater vehicles (AUVs). Improvements in underwater navigation continue to contribute to the expansion of our capability for exploration, scientific discovery and environmental monitoring. Despite the diversity instruments and algorithms employed to improve performance, the fundamental trade-offs involved in estimating position can be understood with a few general stochastic observation models and informationcentric analysis. We present models for the most common oceanographic navigation observations: absolute reference solutions such as long-baseline (LBL) acoustic positioning, relative positioning solutions such as Doppler velocity log (DVL) dead-reckoning and accurate heading such as a fiber optic gyro (FOG) reference. We combine these simple models using the Cramér Rao lower bound to predict both the precision and accuracy of common AUV positioning solutions. To demonstrate the utility of such predictions, we model the uncertainty of current solutions and proposed future implementations. Many current solutions use a combination of LBL positioning and DVL odometry. We show how this approach can predict the performance of such a configuration. Finally, we substantiate the use of this framework by comparing model-based performance predictions to experimental evidence. The field experiment we present is a controlled evaluation of DVL dead-reckoning, using the Jason remotely operated vehicle (ROV). We observe a total error growth that can be approximated as 0.04% of distance traveled.