An innovative artificial intelligence model developed at Arizona State University (ASU) is transforming how scientists understand Mars. This project, led by doctoral student Mirali Purohit from the School of Computing and Augmented Intelligence, has successfully processed approximately 12 million images of the red planet.
Purohit, working within the Ira A. Fulton Schools of Engineering, collaborated with experts to create the Mars Orbital Model (MOMO), a pioneering AI system designed specifically for Martian data. Traditional models struggled with data from Mars due to its unique geological features. MOMO, however, adapts to these challenges by integrating various imaging techniques, providing a comprehensive view of the planet.
The model emerged from a rigorous process of filtering and cleaning an initial dataset of 40 million images, acquired from multiple Mars missions. Unlike previous approaches, MOMO merges different models based on image types, allowing scientists to analyze everything from microscopic features to vast landscapes seamlessly.
Beyond its technical achievements, MOMO marks a significant step towards more accessible planetary science. The Kerner Lab at ASU plans to release both the model and the refined image dataset, enabling researchers globally to engage with Mars studies without the need for custom tools.
Looking forward, Purohit aims to integrate orbital data with rover imagery, further enriching our understanding of Mars’ surface. As she prepares to complete her doctoral studies, her vision includes advancing AI applications in planetary science and eventually exploring Mars firsthand.
