Mars Curiosity acquired artificial intelligence for research

The new Curiosity rover software uses artificial intelligence and allows the rover to independently select targets for scientific research.

Curiosity has been working on Mars for about five years now - it's the most perfect, large and complex planet-planet ever created and sent by humans to another heavenly body. Scientists constantly analyze the images and data sent to them, choosing the next research objective and referring to it. The perfect Curiosity navigation system allows him to partially take on tasks to analyze the environment, obstacles and soil conditions, moving more or less independently. However, more recently, he began to undertake and for the selection of new goals for scientific work.

No, there was no "robot revolt", just scientists working with Curiosity, updated its on-board software. Now, instead of losing precious hours to receive instructions from Earth, the rover can rely on its own "brains". For this purpose, a special addition to the software was downloaded to his computer - the AEGIS (Autonomous Exploration for Gathering Increased Science) system. About him Raymond Francis (Raymond Francis) and his colleagues from the Jet Propulsion Laboratory of NASA, as well as France and Denmark, reported in the journal Science Robotics.

According to the developers, AEGIS was previously tested on the rover Opportunity, and now it is only adapted for the new rover. In the hours when the current tasks are completed, and a new communication session with the Earth is not yet expected, the system includes "robotschenogo" in Curiosity. AEGIS allows the device to look around with on-board cameras, use artificial intelligence to select a suitable target, and then use the ChemCam tool, which shoots it with short but powerful laser pulses and analyzes the composition of the formed plasma. Once the connection to the Earth appears, this data is sent to the scientist.

According to Raymond Francis and his colleagues, earlier, when Curiosity acted randomly, it found itself in dust-naked breeds in only 24 percent of cases. Thanks to AEGIS and its AI, this level reached 94 percent, really "speeding up the collection of scientific data" about the Red Planet.