Time-synchronized acoustic data transmissions can be used to calculate the relative range between underwater vehicles. Relative range, along with other information shared over the communication link, can be used for cooperative navigation. The low bandwidth and latency of the acoustic channel do not allow for a direct implementation of a standard Kalman filter for cooperative navigation. Specifically, there is insufficient bandwidth to share the data needed to accurately propagate the estimation error covariance matrix. Instead, we share only a scalar function of the main diagonal elements of the covariance matrix. We present analysis that suggests this is a reasonable approach and experimental results that demonstrate its efficacy.
This paper presents an system identification algorithm for the longitudinal motion of a streamlined, tail-controlled, miniature autonomous underwater vehicle (AUV). The motion of a rigid body in six degrees of freedom is restricted to the dive plane to derive a fourth order dynamic pitch axis model. Based on the mechanical design of the vehicle, we impose symmetry and buoyancy assumptions that allow further model truncation. Basic dependencies between inherent stability of these truncated models and system parameters such as hydrodynamic coefficients are addressed. For the goal to rapidly design an AUV attitude control system, we experimentally show that it is sufficient to consider a second order dynamic model. We identify model parameters by matching the input and output behavior via least squares. Data collected during field trials validates both the system identification and the utility of a second order model for control design.
A dynamic model is developed for a small, highspeed autonomous underwater vehicle. The vehicle has the unusual property of weighing up to 50 percent more than its displacement, and we seek to characterize requirements for steady-state flight imposed by the vehicles's weight. The process of deriving a dynamic model, which is presented in a tutorial manner, highlights problems that arise when the dynamics of a submersible are expressed in terms of linear and angular momenta and combined with experimentally derived hydrodynamic coefficients
11 November 2012 in Journal of Intelligent & Robotic Systems
We present a fully decentralized motion control algorithm for the coordination of platoons of mobile agents with highly restricted communication capabilities. In order to address very low bandwidth communication between agents and time varying communication network topologies, we utilize a distributed full-state observer onboard each agent. Agent motion and data fusion algorithms are implemented locally by each agent based on the state of the local full-state estimator. Although no separation principle exists between decentralized agent motion control and distributed data fusion in general, we introduce a gradient-based framework in which simultaneously we achieve asymptotic agreement among full-state estimators and convergence of desired agent motion.