Image supplied

How you can help steer the Mars Rover in outer space

NASA’s new tool AI4Mars will help the team of engineers driving the Mars Curiosity Rover to navigate the Red Planet.


Image supplied

NASA is now making it possible for non-professionals to help the team driving the Mars Rover navigate the Red Planet.

You can help to train an artificial intelligence algorithm to guide the Curiosity Rover using an online tool called AI4Mars – Artificial Intelligence For Mars.

AI4Mars enables the user to label features of the terrain from pictures downloaded from outer space.

With this new tool you can help the scientists to identify if what you see is sand, flat bedrock or a plateau, with an option to choose terrain labels.

Citizen science website Zooniverse hosts the AI4Mars tool. It allows you to draw boundaries around terrain and choose one of the four labels.

These “labels” are paramount to sharpening the Martian terrain-classification algorithm aptly named SPOC (Soil Property and Object Classification).

NASA’s Jet Propulsion Laboratory, which is responsible for managing all of the agency’s Mars Rover missions, developed it.

Essentially what SPOC does is label different terrain types and create a map that will help mission team members determine which paths they should take.

The SPOC system is already in place, although it might need further training.

“Hundreds of thousands of examples are needed to train a deep learning algorithm,” according to JPL AI researcher Hiro Ono.

Ono compared this process to the algorithms needed for “self-driving cars”. These are typically trained via numerous images of roads, signs, traffic lights, pedestrians, and other vehicles that may use the road.

Apparently once SPOC is fully up and running, the tool will be able to distinguish between cohesive soil, high rocks, flat bedrock and dangerous sand dunes automatically. It will then send images to Earth that will make it easier to plan the Curiosity’s next moves.

The engineers at JPL are called Rover planners as their job is to plan the Rover’s explorative journey. At the moment it takes up to five hours to plan a drive, as multiple people need to jot down and review hundreds of lines of code.

This task includes in-depth collaboration with the scientists involved.

Geologists, for example, are needed to determine what type of rock the Rover will encounter. They look at whether it will be slippery and the Rover’s wheels may skid, or if there are craters where the wheels could get stuck and damage the vehicle.

The team also has to keep in mind that the Rover’s antenna needs a clear line-of-sight to Earth to receive commands. To determine distance, they use a technique called Visual Odometry which compares camera images to nearby landmarks.

This all takes an immense amount of planning but the development of this new tool will make these tasks a lot easier.

It is not likely that the SPOC will replace the time-intensive work of the planners, but it will give them more time to do other important work such as planning which rock to study next.

This new algorithm could even help the planners for the next Mars mission that NASA has already planned. The new mission, the Perseverance Rover, is due to launch later in 2020.

They are also translating the site from English into languages such as Spanish, Hindi, Japanese and several more so that scientists in these nations also can contribute as well.