Look Mob, No Hands! Automating Mobility Downlink for the Mars Science Laboratory

Now 12 years into its mission, the Mars Science Laboratory’s Curiosity Rover has conducted over 1200 drives, with 83% of those drives successfully reaching their intended destination. No matter success or failure, it is the job of the Mobility downlink chair to assess the performance of all drives that Curiosity executes, examining telemetry to look for anomalous behavior and diagnose the cause of drive faults when the rover does not reach its target. This tactical assessment of drive performance should occur in under 45 minutes, at the end of which the Mobility chair provides a Go or No-Go for rover planning to proceed. During downlink assessments when a drive occurred, the Mobility chair is responsible for reporting on 16 aspects of Mobility performance, only 3 of which require human judgement, meaning the assessment of the other 13 could potentially be fully automated. Automating the downlink assessment and report generation of the Mobility chair addresses two primary challenges faced by the Mars Science Laboratory Mission: reduced funding in extended mission and a desire to increase operational efficiency. This paper reports on the effort made by the authors to automate large parts of the downlink assessment for the Mobility subsystem of MSL, in addition to the results of their effects on the downlink process. The design decisions made for this application of downlink automation are detailed and considerations for future missions are provided for use in the development of their downlink systems to facilitate for reduced need of humans-in-the-loop. Since implementing the introduced automation method, over 80% of the work for the Mobility downlink chair is handled automatically, and assessment times for drives that execute successfully have dropped to under 10 minutes. In the event of a failed drive, the automation method still conducts 80% of the assessment, freeing up over 30 minutes for the Mobility chair to fully investigate the fault and provide an in-depth report, increasing confidence and accuracy in their Go/ No-Go assessment. For a mission still in development, designing their systems to operate with this increased efficiency from day 1 of the mission would pay dividends over the course of the primary mission, and become invaluable in the event of an extended mission.