DUNE: Unified Navigation Environment for the REMUS 100 AUV - Implementation, Simulator Development, and Field Experiments

Abstract

This master thesis presents a new software system for developing control systems on the REMUS 100 autonomous underwater vehicle (AUV). Software written to control autonomous vehicles generally involve many components, and issues such as communication, synchronization, and modularity will arise. Many software frameworks have been written to address these issues. To ease the development, we have implemented an interface between the REMUS vehicle and the open source DUNE framework, leveraging its field-tested software for AUVs. To be able to simulate software written in DUNE, we have implemented an interface to an AUV simulator, AUVSim.

We have used DUNE to control the heading and the altitude of the vehicle. Two altitude controllers have been presented that use range measurements from a Doppler velocity log (DVL) as control input. The auto altitude controller uses the ranges to estimate the altitude of the vehicle. The highpass altitude controller uses an additional estimate of the seabed slope, resulting in a reduced altitude error on large bathymetric variations. Both altitude controllers have been implemented in DUNE and simulated in AUVSim with good results.

A field test was carried out in Hopavågen April 20-23, 2015 to test the two altitude controllers and the heading control via DUNE. The field test showed good result for the auto altitude controller. However, the highpass controller proved to be sensitive to noise in the range measurements from the DVL and did therefore give a noisy control output. Control of heading was not conducted due to an error in the REMUS API. This error was fixed after the field test was complete.

The field tests proved that DUNE was a good fit for controlling the REMUS 100 AUV. The NTNU REMUS 100 can now use DUNE as a development platform, allowing reuse of code from related research activities.