Autonomous docking for the NTNU Otter: Navigation, modelling and control
Background
The Otter is a small unmanned catamaran, propelled by two electrical, fixed thrusters. The vessel is based around the hull, thrusters and power-distribution from a Maritime Robotics Otter, but with sensors and control systems designed at NTNU. This design is based on the LSTS toolchain, with DUNE running onboard the vessel.
Some applications of the Otter include fish tracking (the Fish Otter project), remote sensing and adaptive sampling of e.g. algae or pollution.
For more information on the NTNU Otter, see its web page.
Scope
In order to operate autonomously, the Otter needs to know its position and orientation, which can be determined from its GNSS equipment, consisting of two antennas and receivers to also find the heading, and from the inertial measurement unit (IMU) which provide acceleration and angular rate measurements. Currently, only the position and heading measurements are used. This is sufficient for controlling the Otter, as it is under-actuated, but detailed knowledge of the attitude of the Otter is necessary to e.g. determine what a sensor mounted on the Otter is looking at, to enable geo-referencing of the measurements.
Proposed tasks
- Familiarize with the Otter platform and its sensor suite
- Implement a Kalman filter for navigation based on GNSS and inertial measurements, in matlab and/or in DUNE for running onboard the Otter.
- Test the filter on existing datasets, benchmarking against a “ground truth” estimate from a commercial inertial navigation system.
- Conclude the work in a written report
Possible extensions for the master thesis include
- Docking: in e.g. adaptive sampling, the Otter will collaborate with other unmanned aerial and surface vehicles, as well as satellites, in finding areas of interest that should be investigated. One possibility is that a satellite performs a coarse scan of the Trondheim fjord, and reports a region of interest that the Otter should investigate more closely, to obtain higher-resolution data. This requires the Otter to disembark upon receiving the message from the satellite, as well as docking after completion of the mission, all autonomously. This problem will rely on the navigation solution implemented above, but possibly needs to add other sensors to increase the accuracy close to the dock. Control strategies for disembarking and docking could also be investigated.
- Parameter estimation for optimal control: once the accurate navigation is established, it can be utilized to establish a dynamic model of the Otter based on its measured and estimated states. This has applications in simulations for testing/debugging the Otter control system, planning e.g. how to dock in an optimal manner, model-based control such as model-predictive control for docking and model-based estimation for e.g. estimating the waves and current.
Prerequisites
This is a list of recommended prerequisites, more to signal what it will involve than to be used as a filter on candidates.
- Byggern TTK4155 - Embedded and Industrial Computer Systems Design
- Kalman filtering, from e.g. TTK4115 Linear Systems Theory TTK4250 Sensor Fusion, TTK4190 Guidance and Control of Vehicles
Contact
Contact researcher Kristoffer Gryte for more information.