"Time is precious on Mars," said lead system engineer Raymond Francis in a statement. "AEGIS allows us to make use of time that otherwise wasn't available because we were waiting for someone on Earth to make a decision."

The AEGIS software operates in two different ways: autonomous target selection and autonomous pointing refinement. Basically, these two systems allow the rover to select targets and refine its own laser targeting to analyze samples chosen according to the parameters scientists have selected beforehand. The software has performed at a very high level, exceeding 93 percent accuracy when choosing the correct materials to analyze. According to the paper, the AEGIS autonomous system has "substantially reduced lost time on the mission" and increased the speed of data collection. Before AEGIS was implemented last year, the rover carried out blind targeting, just in case it hit something worthwhile. "Half the time it would just hit soil -- which was also useful, but rock measurements are much more interesting to our scientists," Francis said.

The system has been so useful that NASA is including AEGIS in its upcoming Mars 2020 mission, with an updated version of the imaging system called SuperCam. This new device will have more advanced analysis capabilities to study the crystal structure of rocks.

Excerpt from:

Mars Rover's AI is really good at selecting rocks to analyze - Engadget