Localization or navigation of AUVs using only onboard local sensors, such as a Doppler Velocity Logger (DVL) or Inertial Measurement Unit (IMU), are certain to experience accumulated positioning error. One can, of course, utilize more precise sensors to reduce the rate of accumulated error but uncertainty will eventually grow. Two approaches have been considered in this work:
- AUV and CNA Cooperation: We aim to bound AUV uncertainty using measurements from a Cooperative Navigation Aid (CNA) operating on the surface. This is an MIT autonomous kayak equipped with a WHOI acoustic modem as well as a basic GPS and compass sensor suite. Receiving position measurements from the CNA and estimating the inter-vehicle range allows the AUV to correct its own position estimate and to continue operating longer and more accurately than without this information. On significant complication is that this system uses a single surface vehicle, to minimize expense, and thus is required to be mobile to as to improve positional coverage. Also the system is specifically designed to scale to large AUV fleets in future. [IJRR 2008. FSR 2009. IJRR 2009, ICRA 2011]
- AUV-only Cooperation: Consider an operation in which several AUVs with significantly different performance are operating in the same region - such as a fleet of AUVs doing harbour security or carrying out survey work. If only a single vehicle could surface, aquire GPS position and share this information (using acoustic ranging modems) with the other vehicles their missions could continue without interuption. Alternatively if a single vehicle has a high quality INS it would remain accurately localised for a longer period of time than a low quality vehicle. By sharing this information with the other vehicles in the fleet all vehicles can benefit from the high performance of the expensive vehicle. Current work is considering how the vehicles can share their navigation so that these benefits can be gained - without becoming overcertain [ICRA 2010]
(don't have access to these videos atm)
Video 1 presents the core concept as tested in the Charles River here at MIT. In the centre is the AUV (in this case simulated by a kayak) while to the right is the CNA which provides position and range measurements to the AUV using the WHOI modem. To the left is the dead-reckoned AUV position, which has at that stage hugely drifted off course. Ground truth is via GPS.
Video 2 presents the adaptive control of the surface vehicle, so as to provide optimal position estimates. Both of these videos are screen captures of the shore-side viewer that we use, pMarineViewer, they are part of MOOS-IvP
Video 3 is not technically Cooperative Navigation. Its the result of postprocess of ship (RV Leonardo) to AUV ranges during an AUV malfunction. The ranges were our only information as to the AUV position until it aborted its mission and surfaced.
Video 4 is the first proper and online test of the full AUV and CNA Cooperative Navigation system. This uses OceanServer's Iver2. The AUV surfaces twice during the mission, at the ends of the tracks and in both cases the CN system was significantly closer to the surface position than the onboard dead-reckoning system estimated.