Enabling autonomy in commercial-class lunar missions

Early Lunar micro-rover missions will be short in duration and have constrained downlink capacity. To maximize the scientific return of these missions, Mission Control is developing technologies to autonomously classify geological features and detect novel features in rover camera imagery, which can be used to support intelligent decision-making for prioritizing data for downlink and instrument targeting. In a recently completed concept study, a trade-off analysis and performance evaluation were conducted for the terrain classifier and novelty detector algorithms across multiple datasets. The terrain classifier developed achieved accuracies of 77%-86% and Intersection over Union (IoU) scores of 0.667-0.680 across 10 different terrain type, on 3 distinct data sets (totalling 928 images), demonstrating the robustness of the approach to varying illumination conditions. In ongoing work, a comprehensive Lunar analogue dataset is being developed to continue prototyping, and the algorithms are being developed on an embedded processor for a flight demonstration opportunity.