Mars Science Laboratory R13 Mobility Regression Testing

Now in its 10th year of Operation on Mars, NASA’sCuriosity Rover is preparing to undergo a software update, onethat has been six years in the making. In order to qualify thisupdate for use on Mars, the flight software first needs to be runthrough a regression test campaign here on Earth. This paperdetails the regression testing efforts of the Mobility team as theyevaluated the performance of the rover’s driving capabilitiesunder this new Flight Software Upgrade - known as R13. Thistest campaign was conducted to ensure that all of the rover’smobility capabilities behave as expected, especially to check theparts that are “untouched” by R13 have not regressed inperformance, hence the term regression testing. Of 75 mobilitycommands, Curiosity has 18 commands that initiate motion andeach has a number of arguments and options that cancompletely change the rover’s traverse behavior. This leads to avast and complex problem space that is impossible toexhaustively explore, necessitating a regression test campaignthat targets key subsets of commands and arguments in orderto create the confidence needed to approve R13 for flight. Toconduct this testing, the Mobility Team spent 10 days in the JetPropulsion Laboratory’s (JPL) Mars Yard commanding theVehicle System Test Bed (VSTB) – Curiosity’s Earthboundtwin. The nominal performance for 11 of Curiosity’s motioninducingcommands was evaluated in addition to the FaultProtection behavior for 29 off-nominal cases that could occur inrelation to mobility activities. The off-nominal testing requiredcommanding the rover in a way that the Operations Team neverintentionally would on Mars, purposefully stalling motors,shutting down the rover mid-drive, and exceeding limits that arehopefully never reached in flight. The Mobility team’s 10 shifts,the most of any subsystem, created two significant challengesconsisting of staffing and data management. Unlike most otherspacecraft testbeds that spend their days in clean rooms,Curiosity’s drives around the Mars Yard and requires at leasttwo testers to operate. One of which commands the rover, andthe other who moves the umbilical cable that provides powerand communication between the command station and vehicle –a physically demanding task in the Southern California summerheat – leading to the often addition of a third tester to lightenthe load. After testing concluded, data from the tests wasanalyzed to confirm that the rover behaved as expected, and ifnot, what further testing or documentation would be requiredto proceed with approval for R13. Finally, the Mobility Teampresented their test results to the regression review board andthe test campaign was approved, bringing R13 one step closer torunning on Mars.