Autonomous Submersible Robot: State Estimation System

The aim of this project was to develop the state estimation system for an autonomous underwater vehicle called Kambara. This project took place at the Research School of Information Sciences in cooperation with the Department of Engineering at the ANU. The purpose of the state estimation system was the filtering of data from a triaxial accelerometer, rate gyroscope, depth sensor and compass/inclinometer to track the position, orientation, linear velocity and angular velocity of the Kambara. To accomplish this, an extended Kalman filtering algorithm was proposed. Nonlinear models were derived for both the sensors and the dynamics of the Kambara. Device drivers for the underlying hardware architecture were developed so that data could be logged from the sensors onto computer. Static testing of the sensors has begun for the sake of modelling noise processes and temperature sensitivities. The performance enhancements of prefiltering the sensor outputs with analog filters were investigated. Plenty of scope is left for future work on the state estimation system. Further characterisation of the sensors will be required. Coding of the Kalman filter algorithm and parts of the sensorsampling algorithm are still required. Eventually, the state estimator will be implemented onto the Kambara computer and the whole system tested and tuned. All the while, hydrodynamic and geometric parameters need to be estimated for implementation of the dynamic model.