Present: Rich, Rob, Frederic, Conor and Kanna

• A test AMC stack for the autonomy computer is now finally available. It is trex1.shore.mbari.org and has a standard Fedora/RH distribution. Quite a bit of testing with TREX and the EUROPA codebase has been accomplished over the last week since this came online.
• However due to the current RH gcc version, EUROPA code had be modified to compile with stricter gcc 4.1 which added quite a bit of work for Conor.
• The next steps in cloning this machine to go onboard the AUV's bubble would be:
  • to test the process of building an image offline on a faster machine and dropping it into trex1 for testing.
  • start performance tests to determine how TREX is doing. Current indications on the Mac appears to be that less that 2% of CPU time is being taken over by an optimized version of TREX, primarily for synchronization and database updates.
• We also have a QNX test environment in foobar.shore.mbari.org ready to go.
• There appears to be a problem in the socket code when testing on vehicle which appears to be timing related. Frederic who implemented the original code will be evaluating the code this week.
• There appears to be no trace of state publisher when using the vehicle. Rich and Conor are planning to investigate.
• An issue of vehicle dynamics and layered control came up to see what the vehicle would do when a requested Yo-Yo behavior is not doable because of dynamics
  • Rob will investigate by checking the layered control code
• An associated issue (which will impact the autonomy task later) is how Frederic's clustering algorithm is to work with the behavior which needs to use it to initiate a sample. The latter is being written by Rob who will coordinate with Frederic to see how these two software modules need to work together.