Autonomous Regolith Extraction Using Real-Time Diagnostics and Dynamic Plan Execution for 1 Meter Class Interplanetary Rotary-Percussive Drills

Abstract

As remotely operated interplanetary drills probe deeper strata, autonomous health monitoring and fault recovery is required to accommodate communication delays that well exceed the time required for such systems to become stuck. This paper will discuss work on autonomous diagnostics and fault recovery software coded in the plan execution interchange language (PLEXIL). By caching sensor and motor data as it is collected in real-time, the software evaluates the drill’s current operational state and classifies the state as either nominal or one of a number of uniquely defined off-nominal fault states. Each drill fault requires its own unique recovery procedure, which is executed by the software upon detection. The software verifies the success of the recovery based on a series of goal conditions that are dependent on the fault being recovered from. If the recovery attempt succeeds, the drilling operation is resumed; otherwise, the recovery procedure is reattempted as many times as needed until the goal conditions are achieved. The software has been tested on 3 different rotary-percussive, Mars-capable robotic drills; the latest of which is the Atacama rover astrobiology drilling studies (ARADS) drill. Tests on the ARADS and other drills occurred in a variety of Mars analogue environments, including: the arid Atacama Desert, the permafrost encrusted Devon Island, and the anaerobic bacteria habitats around the Rio Tinto. An overview of the autonomous diagnostic and recovery system, with analyses of how they performed in these Mars-like conditions, are given.