Skip Navigation LinksAutonomy

Du er her: Skip Navigation Linksffi.no / no / Forskningen / Avdeling Maritime systemer / HUGIN / Autonomy on HUGIN
Forskningen
Autonomy on HUGIN
Skriv ut
Tips venn
Facebook
Twitter
02.02.2011
Linje
By giving an AUV a flexible system for decisional autonomy it will be able to handle unexpected situations in a robust at mission-efficient manner.

If parts of the vehicle malfunction, some mission goals may still be obtainable, while others have to be discarded. Furthermore, an advanced autonomy system will open up new kinds of missions, such as missions that involve using information gathered at the beginning at a later stage. An example is to return to interesting objects discovered with a side scanning or synthetic aperture sonar with an optical camera, in order to further investigate and identify them.

 

HUGIN Autonomy Layer – HAL

FFI’s HUGIN has been equipped with a software framework for advanced autonomy called the HUGIN Autonomy Layer, or HAL. HAL enables autonomous features to be implemented in HUGIN, increasing the level of autonomy step by step. The framework has a two-layered structure with a deliberate/executive layer and a reactive layer. The deliberate/executive layer is responsible for planning and thinking ahead, while a reactive layer is responsible for quickly reacting to sensor input, in a manner that meets the goals set by the deliberate/executive layer. If the reactive layer encounters situations (i.e. a set of sensor inputs) that it does not know how to handle, it will notify the deliberate/executive layer. This layer will then analyze the new situation and replan (parts of) the mission if it is deemed necessary. 
 
The structure of the HUGIN Autonomy Layer
 
Execution of a preplanned mission plan has been implemented in HAL. In addition, four autonomous features have been implemented and tested: The autonomous GPS fix, the autonomous CTD fix, layer following and adaptive linespacing.
 

Autonomous GPS fix

The autonomous GPS fix feature allows HUGIN to pause the current mission and go to the surface to get a position fix from the GPS. The reactive layer monitors HUGINS navigational uncertainty, and notifies the deliberate layer when the uncertainty is too high. The deliberate layer will then decide if a GPS fix is possible (e.g. HUGIN should not be too deep), and if so pause the mission and perform the fix. 

Navigational uncertainty and depth during an autonomous GPS fix
 
 

CTD fix

The CTD (Conductivity, Temperature and Depth) fix aims to avoid a large error in HUGIN’s estimate of the sound velocity profile in the water column around it. As the most common payload sensors aboard HUGIN (such as sonar and multibeam echo sounder) is uses sound and echo, sensor performance is dependent on a good estimate of the sound velocity.
Before a mission, the sound velocity profile of the operating area will typically be uploaded to HUGIN. If the reactive layer senses that the actual sound velocity differs from the stored velocity profile, it will initiate notify the deliberate/executive layer. This layer may then pause the mission and initiate the CTD fix. The vehicle will measure the CTD both above and below HUGIN’s current depth, storing the measured data and the calculated sound velocity into a new CTD and sound velocity profile.


 

Adaptive line spacing

When HUGIN maps or searches an area using a synthetic aperture sonar (SAS), it will typically move in a lawnmower pattern.  The distance between the lines of this survey pattern must be short enough to allow full coverage without holes. At the same time, the number of survey lines should be minimized to reduce the time spent surveying the area. The optimal distance between lines is the distance that gives full coverage (with the desired amount of overlap), but not more that the desired overlap. This distance is dependent on SAS performance, which may vary with environmental parameters that are hard to estimate.
By autonomously measuring SAS performance in real time during a survey line, the distance to the next line can be planned to achieve the desired overlap of coverage, using the SAS range of the previous line as an estimate of the range of the next line (may not always be correct).  This feature will both increase the robustness of survey performance, and reduce the planning workload of a HUGIN operator.
 
 

Layer following

When in layer following mode, HUGIN will attempt to follow a layer in the water column. A plan of the horizontal movement will have been given to HUGIN, and it will adjust its depth to stay within the layer.

Illustration
 
HUGIN is currently able to follow layers in the water density, which depends on water salinity, temperature and depth. However, it is also possible implement autonomous layer following of different kinds of layers, such as temperature, plankton density or chemical concentration of an interesting substance.
A related problem to layer following is gradient following, where the vehicle follows the gradient of an interesting field (such as concentration of hydrocarbons in the water, if the purpose is to find a leak or source).
 
 
© Copyright Forsvarets Forskningsinstitutt