Aegis Autonomous Targeting for Chemcam on Msl: Results from the First 220 Sols of Routine Science Operations

Introduction: The Autonomous Exploration Gathering for Increased Science (AEGIS) software suite [1] was uploaded to the Mars Science Laboratory (MSL) mission’s Curiosity rover in late 2015 and tested on the surface of Mars over several months [2]. On 6 May 2016, the system was approved for use by the MSL science operations team, and first ran in routine science operations on sol 1343. This abstract gives an overview of initial use of the system, results, and its effects on ChemCam and MSL operations and science productivity. Overview of the AEGIS system: AEGIS is an intelligent targeting system allowing the rover to select its own targets for the ChemCam instrument at times during the mission when Earth cannot readily be in the loop – such as immediately after a drive. It is used in two ways – by detecting and selecting targets in NavCam stereo images, then measuring them with ChemCam, or by detecting features in ChemCam’s own Remote Micro-Imager context camera, and recentering on them as a form of pointing refinement to account for pointing error and uncertainty. In this abstract, we focus on the use of the autonomous postdrive target selection with NavCam source images. Use since May 2016: The rollout of AEGIS to the science team consisted of an initial set of reusable activities intended to be run post-drive, with AEGIS’ computer vision and target ranking parameters tuned to find patches of bedrock. After several weeks of experience, this ‘scene profile’ was further adjusted to adapt to the materials present at that point in the rover’s traverse, improving the already high rate of finding and measuring bedrock targets. We refer to these as ‘version 1 and 2’ of the initial scene profile. The AEGIS software has run on NavCam images 36 times from May through December 2016 using versions 1 & 2 of the initial scene profile, selecting 40 rock targets (some runs are configured to select two rock targets, when time and resources allow). The first 10 runs, using version 1, returned outcrop targets in four cases; two more targets were loose float rocks, with the remaining four being clusters of small rock pieces or patches of soil. With adjustment of the target parameters to version 2, the performance in selecting outcrop significantly increased. From sol 1400 to 1526, 26 runs returned 28 outcrop targets, and two float rocks which were representative of nearby bedrock. This increase from 40% to 93% in the rate of outcrop reflects the adaptability of the system, and the ability of the operations team to adjust parameters and adapt to changing terrain. AEGIS has increased the pace of ChemCam activity – there has been a 10% increase in the number of ChemCam science sequences executed on-board per time, and a corresponding increase in the rate of LIBS shots, since the system went into use (Fig.1). Further use of AEGIS along with continued target selection optimization will only further enhance the return of science.