October 17, 2004
Guanaco Camp, Atacama Desert, Chile
Agenda
• Continue autonomous traverse
• Continue fault recovery experiments
Status and Progress
Continue Autonomous Navigation Experiments. Today Zoë passed the 50
kilometer mark navigating autonomous. This distance was the goal we set
for this season as something achievable with effort that would ensure
that Zoë and the autonomous navigation software experienced enough
terrain to reveal the strengths and weaknesses of our designs and
algorithms. This total traverse distance is the sum of over 250
individual traverses, some short, some quite long, each initiated to by
a single command sequence uploaded to the rover. Some commands were
fairly simple, go to a specific location in the distance, some where
more complex involving multiple waypoints and science actions. The
command was decomposed by onboard planners into shorter actions, which
are commanded by the rover executive to navigation and instrument
modules. The farthest Zoë ran autonomously was 3.3 kilometers but on
average a traverse would terminate after just over 200 meters. We were
particularly pleased that our rover and navigation approach was able to
reach beyond 1 kilometer on 9 occasions. (For what it’s worth, there
were 2 addition traverses that exceeded 995 meters.) We still have a
couple more days of testing.
Continued rover executive experiments. The rover executive is the
software module that accepts plans from one of the mission planners,
tracks progress against the plan, doles out commands to other software
modules, and reacts to faults detected during operation. We have a very
simple rover executive used only for autonomous navigation but this
year in efforts to fully automate the science investigation (a goal for
next year) we have been developing and testing a rover executive that
is capable of interpreting complex plans of navigation and science
goals, monitoring the rover, sensors, instruments, power,
communication, and all the other software modules, and guiding all
onboard activities. This is a difficult software element to design and
even more challenging to test and prove reliable. In conjunction with
our autonomy experiments we have been exercising the rover executive
making sure that it can operate under nominal conditions and we have
been intentionally introducing faults to determine whether it correctly
recovers from anomalous conditions. Zoë is quite robust against the
types of faults that are most common, for example, failing to find a
clear path ahead, and we are now working on the more unusual situations
that still cause the rover to halt and await assistance (or
occasionally carry-on oblivious to the problem).
Upcoming
• More autonomous traverse and fault recovery experiments
• Conduct mission planner validation tests
Weather
Dry, dry, dry, 10% humidity.
Quote of the Day
"Well, that’s not going to work."
For more information on Life in the Atacama including images, movies,
and field reports see: http://www.frc.ri.cmu.edu/atacama
