Monterey Bay Aquarium Research Institute
Autonomy
7 hours of underwater adaptation
Shore-side Mixed Initiative AUV commanding to follow Fronts


May 7th 2009 :

We have been working with John Ryan to augment onboard autonomy with the ability of scientists to command the vehicle from shore with science based objectives and doing so opportunistically.  In this the objective was to demonstrate Mixed-Initiative (human in the loop) control to follow an ocean front. Our initial demonstration was designed with the following steps:

  • The vehicle does an initial "mapping" transect while collecting data samples for analysis
  • Using a compact representation, the vehicle sends data to shore via Iridium
  • On shore, tools help unpack this data for analysis by a scientist
  • The vehicle awaits high-level goals based on the shore-side analysis, which in turn retargets its mission objective
  • The new goals are integrated onboard and the vehicle synthesizes new plans to its mission. In this case, the objective was to home in on the hot-spot of a front and track it repeated while the front is moving. It does so by repeatedly :
    • searching for a temperature anomaly at a given depth
    • as soon as this hot-spot is identified to do a short transect around the point where the temperature threshold was crossed
As AUVs become more persistent in the water column, such techniques will be necessary to allow retargeting from shore to follow new objectives where substantially more data and computational power will be at the disposal of scientists while augmenting her/his capacity to have an extension of the human senses in the unstructured ocean.

For the May 7th demonstration, we selected a south-north transect in the northern Monterrey Bay (see figure on right) going past the M0 mooring especially since  data collected by the vehicle can be correlated with the mooring. The initial transect was approximately 11km long and roughly 2 hours in duration. The vehicle surfaced around the northern end of the transect while polling its Iridium modem for new messages. Data received from the vehicle helped Ryan identify a number of weak fronts as illustrated below. A threshold of 10.8รด C around the depth of 10m was chosen as the front to track.

                                                                                                                                                                        Image courtesy: John Ryan
 
The new objective was sent to the vehicle on receiving which the vehicle proceeded to track as shown in the 2D view below. Along the way the vehicle found two different (weak) fronts which it attempted to map. As they moved out of the vehicles transect the vehicle finally latched onto the front initially targeted. Total mission duration was in excess of 6 hours with 3. The following images show what was collected during the mission with a movie showing the vehicle path of the entirety of the mission.

Movie of the mission :

Last updated: May. 12, 2009