Scientists from NASA want everyone to help navigate the surface on Mars, in order to ultimately facilitate the movement of devices like Curiosity.
The agency has an algorithm called SPOC for determining soil properties and classifying objects that marks various types of Mars rafts, including boulders or sand. The tool is used to create maps that the rover can use when maneuvering, but the system needs refinement, and this requires the input of a huge amount of data.
“Typically, hundreds of thousands of examples are needed to teach the deep learning algorithm,” says Hiro Ono of the Jet Propulsion Laboratory. According to him, the algorithms for unmanned vehicles are trained using numerous images of roads, signs, traffic lights, pedestrians and other vehicles. A similar scheme should be carried out for the Martian instrument, and everyone can help in this.
Third-party users can use the AI4Mars tool to draw borders around objects on the ground and mark them as sand, soil, rock, or large rocks. This will help teach SPOC to distinguish terrain.