Master of Engineering Science
Electrical and Computer Engineering
The CheMin (Chemistry and Mineralogy) instrument on the Curiosity rover has provided a rich set of X-ray diffraction (XRD) patterns from Martian rocks and regolith. These XRD patterns have allowed geologists to make exciting new discoveries about the mineralogy and the geological history of Mars. These discoveries pave the way for further Martian exploration and provide a deeper understanding of Martian geology. The Curiosity rover is very slow by design, travelling at about 4 cm/s. New, faster rovers are being developed to increase scientific throughput and exploration. XRD is valuable for future missions as it can produce new discov- eries and be used to find interesting locations for further investigation. However, the length of time it takes for XRD data to be sent to Earth and analyzed by an expert is a bottleneck that lim- its the usefulness for these types of instruments. Instead, automatic analysis of XRD patterns onboard the rover could be used to identify minerals and inform navigation when looking for scientifically interesting materials. To this end, an XRD dataset of 42,000 XRD patterns with up to 15 out of 84 possible minerals in each pattern was created using extensive physics-based augmentation. Several convolutional neural networks (CNN) with different architectures were trained and compared on this dataset. The best model was then tested on 45 CheMin XRD pat- terns and 7 Martian analogue patterns, obtaining an F1-score of 60%, and 38%, respectively. These scores correspond to performance of 39% and 15% above chance level, representing a good first step to the problem, given the difficulty of the dataset. Further, the training dataset and insights from this model provide a useful starting point for future work on the topic.
Summary for Lay Audience
NASA’s Curiosity rover is equipped with a CheMin (Chemistry and Mineralogy) instrument that scans rock and soil samples using a technique called X-ray diffraction (XRD). XRD is a technique where a beam of X-rays is shot into a sample and records the angles and intensities of the X-rays that are diffracted by the sample. Because of the way minerals are arranged at the atomic level, each mineral will produce a unique pattern that can be used as a “fingerprint” for identifying that mineral. The CheMin instrument has been used to produce many XRD patterns and has sent them back to Earth to be analysed by expert mineralogists. These have provided great value for learning about the geology and history of Mars. The downside of the CheMin instrument is that it takes several days to make a reading and even longer to send that sample to Earth to be analysed. This was not a problem for the Curiosity rover, since it was designed to be slow and reliable. However, future rovers are being designed to move faster and take more scientific measurements. XRD could continue to be a valuable asset on future rovers, and it could even be used to help find interesting sites for the rover to explore. The problem is that sending the sample to Earth to be analysed by experts is a time consuming process that creates a bottleneck. One potential solution is to have the rover analyze the XRD patterns autonomously, so that human pilots, or even the rover itself, could make better and faster decisions about where to collect further samples based on those results. In this work, methods are investigated for automating the analysis of these XRD patterns of Martian rocks. To do so, first a large training dataset is created, comprised of many Martian-like XRD patterns, using a method that simulates how these patterns might appear on Mars. Using this large dataset, an artificial intelligence (AI) model is trained to predict which minerals are contained in each XRD pattern. After that, this AI was tested on 45 real Martian XRD patterns from the CheMin instrument to see how well it performed on real data. The AI was also tested on 7 patterns that were measured in a lab on Earth of Martian meteorites and samples that are similar to those that might be seen on Mars. The model did not perform perfectly, but it was a good start towards building truly autonomous XRD analysis on Mars.
Tambakis, Luke, "Towards Automated Mineral Identification in Martian rocks from X-ray Diffraction Patterns" (2023). Electronic Thesis and Dissertation Repository. 9630.