The amount of data NASA has about Mars is absolutely enormous and increasing every day. In fact, the Mars Reconnaissance Orbiter has been circling Mars for the last 15 years now and every day it sends back invaluable images that researchers analyze for potential landing sites for rovers.
Scientists are particularly interested in the photos of craters since they can provide a window into the planet’s history. AI is assisting researchers to spot these craters in the images, but the AI system requires lots of training data to further hone the algorithm used to identify these craters. Let’s take a closer look at how such an AI system works and how Mindy Support can be useful in helping to train the machine learning algorithm.
The Mars Reconnaissance Orbiter is equipped with three cameras. One of them is called Context, which is a low-resolution grayscale camera and another is called HiRISE, which uses the largest reflecting telescope ever sent into deep space to produce images with resolutions about three times higher than the images used on Google Maps. These cameras produce thousands of images which are later fed into an AI system that is used for analysis purposes. The algorithm used for detecting craters was trained with about 7,000 images of Mars, some of which contained previously discovered craters.
These and any future images used for training the machine learning algorithm need to be annotated with both labeling and semantic segmentation so the AI system can better identify future creators. Since this is a very time-consuming and low-level task, it would not make sense to perform such data annotation work in-house. Mindy Support can assemble a team of qualified data annotators who will do all of this work for you while keeping costs under control.
In addition to saving researchers a lot of time, it helps them identify craters that could go unnoticed to the human eye. The more recent craters on Mars are small and may only be a few feet across, causing them to appear in dark pixelated blotches. It would be almost impossible for a human researcher to make any sense of such an image.
However, if the AI system comes across such a dark patch and compares it with other dark patches of confirmed craters it could alert researchers that it has discovered something. Interestingly enough, even if the AI system spots a crater, scientists do not simply take the system at its word, so to speak. They send the Mars orbiter to confirm the crater’s existence.
This past summer, the AI system discovered a crater on Mars for the first time. This is the first time AI was used to discover a crater on another planet. This new technology could significantly speed up the discovery of craters on Mars and other planets without sacrificing the accuracy. Over the past 15 years, scientists had to examine each image manually to discover new craters, which would take about 45 minutes per image. The AI system could do that same work in less than 5 seconds.
In addition to this, the craters could also teach scientists what lies beneath the surface as well. For example, about ten years ago, the Mars orbiter spotted a crater that exposed some subsurface water ice. Researchers were able to study the exposed ice and how it disappeared over time to get a better sense of how ice is distributed across the surface of the entire planet. However, additional such discoveries will need to be made to provide some definitive conclusions on the possibility of life on Mars and its history.