NASA’s Mars Rover Drivers Need Your Help

Using an on-line device to label Martian terrain types, you can train an synthetic intelligence algorithm that could improve the way engineers manual the Curiosity rover.

You may possibly be able to aid NASA’s Curiosity rover drivers greater navigate Mars. Using the on-line tool AI4Mars to label terrain features in shots downloaded from the Red Earth, you can train an synthetic intelligence algorithm to mechanically browse the landscape.

Is that a big rock to the remaining? Could it be sand? Or maybe it’s awesome, flat bedrock. AI4Mars, which is hosted on the citizen science web site Zooniverse, lets you attract boundaries around terrain and pick out a person of four labels. People labels are critical to sharpening the Martian terrain-classification algorithm called SPOC (Soil Residence and Item Classification).

A few images from the device named AI4Mars exhibit different kinds of Martian terrain as seen by NASA’s Curiosity rover. By drawing borders around terrain features and assigning a person of four labels to them, you can aid train an algorithm that will mechanically identify terrain types for Curiosity’s rover planners. Credit history: NASA/JPL-Caltec

Produced at NASA’s Jet Propulsion Laboratory, which has managed all of the agency’s Mars rover missions, SPOC labels several terrain types, building a visual map that assists mission staff customers ascertain which paths to acquire. SPOC is now in use, but the procedure could use even more coaching.

“Typically, hundreds of hundreds of illustrations are wanted to train a deep understanding algorithm,” mentioned Hiro Ono, an AI researcher at JPL. “Algorithms for self-driving cars and trucks, for instance, are skilled with many images of streets, indications, targeted traffic lights, pedestrians and other vehicles. Other general public datasets for deep understanding comprise people, animals and structures – but no Martian landscapes.”

When entirely up to velocity, SPOC will be able to mechanically distinguish among cohesive soil, high rocks, flat bedrock and dangerous sand dunes, sending images to Earth that will make it much easier to program Curiosity’s next moves.

“In the long run, we hope this algorithm can become accurate ample to do other beneficial tasks, like predicting how probably a rover’s wheels are to slip on different surfaces,” Ono mentioned.

The Career of Rover Planners

JPL engineers named rover planners may possibly gain the most from a greater-skilled SPOC. They are accountable for Curiosity’s each go, regardless of whether it’s taking a selfie, trickling pulverized samples into the rover’s body to be analyzed or driving from a person location to the next.

It can acquire four to 5 several hours to function out a generate (which is now carried out almost), requiring numerous people to create and review hundreds of strains of code. The job consists of substantial collaboration with researchers as very well: Geologists assess the terrain to forecast regardless of whether Curiosity’s wheels could slip, be damaged by sharp rocks or get caught in sand, which trapped each the Spirit and Opportunity rovers.

Planners also take into consideration which way the rover will be pointed at the finish of a generate, considering the fact that its high-gain antenna needs a apparent line of sight to Earth to obtain commands. And they try to anticipate shadows slipping across the terrain through a generate, which can interfere with how Curiosity decides distance. (The rover uses a procedure named visual odometry, evaluating camera images to nearby landmarks.)

How AI Could Support

SPOC will not substitute the difficult, time-intense function of rover planners. But it can free of charge them to concentrate on other factors of their task, like discussing with researchers which rocks to research next.

“It’s our task to determine out how to safely and securely get the mission’s science,” mentioned Stephanie Oij, a person of the JPL rover planners concerned in AI4Mars. “Automatically building terrain labels would help you save us time and aid us be more productive.”

The gains of a smarter algorithm would extend to planners on NASA’s next Mars mission, the Perseverance rover, which launches this summer time. But 1st, an archive of labeled images is wanted. Much more than 8,000 Curiosity images have been uploaded to the AI4Mars web site so far, giving a good deal of fodder for the algorithm. Ono hopes to increase images from Spirit and Chance in the long run. In the meantime, JPL volunteers are translating the web site so that members who speak Spanish, Hindi, Japanese and several other languages can lead as very well.

Resource: JPL