- Particle Filter based Nonlinear Observer for AUV Navigation
During submerged operation, in the absence of GPS, an AUV is forced to use dead-reckoning to estimate its position and attitude from the vehicle's inertial measurement unit, compass, and doppler velocity logger. Extended Kalman Filtering (EKF) which is the mainstay of most navigation systems uses linear and Gaussian assumptions that are not always correct about the nature of the plant being estimated. We have developed a particle filter based observer, Measurement Assisted Partial Resampling (MAPR) filter to improve navigation accuracy. Unlike the EKF, the particle filter makes continuous approximations of the state error. The result is improved accuracy over extended periods. The particle filter has an additional advantage in that it readily allows incorporation of bathymetric data for improved localization and mapping (SLAM) which is required for survey roles.
- Coupled 6-DOF Control for Autonomous Underwater Vehicles
Optimal guidance control is highly desirable to allow an AUV to manoeuvre reliably in cluttered and hazardous environments, and to accurately position itself in the presence of disturbances caused by currents and waves. Mathematical models of AUVs are highly nonlinear and coupled, which can lead to parameter uncertainty within the control systems that are designed based on simplified versions of these models. Sliding-mode control has been applied to underwater vehicles in several forms, including both fully and partially decoupled DOF control; however, none of these forms has demonstrated strong reliability in practice. Our objective is to develop a fully coupled robust variable structure controller that can handle parameter uncertainty and control all 6-DOF simultaneously and smoothly. This strategy enables the controller to anticipate the reaction of the system to control inputs for all DOFs, and therefore produce a high level of accuracy with respect to the positioning and trajectory of the vehicle especially when it executes manoeuvres comprising simultaneous motion about multiple DOFs.
- Virtual Reality Autonomous Vehicle Simulator
We are developing an AUV simulator to allow realistic testing of the guidance control and navigation components in the vehicle while subjecting the AUV to disturbances and other synthesised sensor data. The simulator provides a Matlab Simulink interface for modelling the vehicleís propulsion, control, navigation, sonar/video sensors, and communication systems. The simulator architecture allows for hardware components such as the vehicleís host computer, motors, etc., to be substituted in place of software models for hardware-in-the-loop testing within the virtual reality environment.
- Prof. Neil Bose, AMC - National Centre for Maritime Engineering and Hydrodynamics
- Assoc. Prof. Colin Kestell, Mechanical Engineering, The University of Adelaide
- Dr. Stephen Grainger, Mechanical Engineering, The University of Adelaide
Joining the group
Opportunities exist for both collaboration and postgraduate research. For more information, please contact the research group leader A/Prof Karl Sammut.
We would be happy to provide more information about the School's research programs, the opportunities for higher degree study and scholarship information. For more information, please contact the research group leader A/Prof Karl Sammut.