Fig. 1
The concept lander, perhaps the most complicated structure ever created using generative design, was unveiled at Autodesk University on November 13th in Las Vegas.

NASA Uses Generative Design on Next-Generation Interplanetary Lander

Autodesk software lets NASA designers balance several performance goals against mission constraints while reaping the benefits of cloud computing and machine intelligence.

Over the last decade, NASA has sent several orbiters, landers, and rovers to Mars with great hope that one of them would discover signs of life on the red planet. So far, no luck. Though they still have their fingers crossed, some engineers and scientists at NASA now think the best chance of finding life, or signs of it, lie on the moons of Jupiter and Saturn. But although the trip to Mars is long and arduous at 35 million miles, the distance to these gas giants’ moon is an order of magnitude farther, with Jupiter roughly 365 miles from Earth and Saturn out at about 756 million miles. The risks are much greater as well.

Getting safely to these planets, and then landing an exploratory rover or sensor-studded probe on one of the moons, presents far greater design and engineering challenges. To meet those challenges, many of which are unknown, NASA’s Jet Propulsion Lab (JPL) is exploring the use of generative design in a multi-year collaborative research project with Autodesk, the company that coined the term.

Mark Davis, senior director of industry research at Autodesk, was part of the team that talked with JPL about a collaboration. “NASA was clear it wasn’t interested in incremental gains,” Davis recalls. “If we could only improve performance by 10%, they basically weren’t interested. If we could deliver software tools to help them achieve a performance improvement of 30% or more, then we had their attention.”

The interplanetary lander was carefully assembled by project team members at Autodesk’s technology center in San Francisco.

To get to one of moons of Jupiter or Saturn, then land on it and explore the surroundings, a lander needs to operate practically autonomously in temperatures far below zero and withstand radiation levels thousands of times greater than those on Earth. At the same time, it must be light enough that a rocket can carry it all the way there.

In space exploration, weight at liftoff is a critical consideration. Every pound that can be cut from the structural payload lets engineers increase the scientific payload of sensors and instruments, or provide more protection against radiation, or add more performance in terms of batteries and expendables.

In some industries, it’s considered a good thing to “fail fast” or get to a “minimum viable product” as quickly as possible, then incrementally improve it. But in space exploration, failures are costly. Missions often have only one shot at success and there are few, if any, viable backup plans. So, teams at JPL are careful when considering new processes. They stick with what has proven to work, such as titanium and aluminum—structural metals they know will hold up in the harsh conditions of space—and mission-proven manufacturing processes such as CNC machining.

That being said, they also need to explore new technologies, lest they risk being made obsolete by other companies. It’s always a balancing act between what’s proven, what’s possible, and the risks.

The team carefully connects the CNC-milled legs to body of the lander.

Within the company, JPL’s Atelier division is charged with examining new approaches and processes, then recommending the ones that hold promise to teams working on specific missions. “They carefully infuse new technology into NASA processes,” says Karl Willis, Autodesk’s technology lead on the project. “They know they have to explore new ways to do things while minimizing risk.”

Generative design is a relatively new approach to the design process. It leverages machine intelligence and cloud computing to quickly generate a broad set of designs that fit within the specific constraints set by mission engineers. It lets design teams explore a larger group of alternative approaches while still being bound by manufacturing and performance requirements dictated by the team or environment.

The interplanetary lander is tentatively designed to explore the moons of Jupiter and Saturn where it might find life or signs of it.

A commercial form of generative design is available today in Fusion 360, Autodesk’s product development platform. Davis and his team in Autodesk’s Office of the CTO continue to develop conceptually advanced versions of the cloud-based software for use in experimental capacities, such as with JPL. “We had developed an earlier custom version of the software for high-performance race cars that let us help customers solve for multiple constraints at once,” Davis says. “We then applied it to the problems JPL needed to consider.

“We took software developed to help race teams solve high-level suspension problems on Formula One race cars and applied new requirements for structural constraints critical to space exploration. This gave us a chance to push the software’s capabilities even further and work with customers solve even larger and more complicated problems.”

Generative design had no problem combining 3D printing, CNC milling, and casting into the manufacturing mix it took to fabricate different parts of the lander.

Generative design is often associated with 3D printing, also known as additive manufacturing, which is well-suited for the complex, organic-looking shapes the software creates based on user specifications. But the software also lets users set constraints for their other manufacturing processes.

“We can now help customers solve for multiple manufacturing constraints simultaneously, which factors in CNC machining and casting options, as well as 3D printing,” explains Davis. And although other software programs can optimize a single part for stiffness and weight, generative design lets design teams consider alternate strategies and produce an entire array of viable solutions, rather than just a single optimized version.

For the lander project, the JPL team explored generative design for several structural components, including the internal framework that securely holds the scientific instruments and the external assembly that connects the lander legs to the main payload box. Using generative design, the team was able to reduce the mass of the external structure by 35% compared with the baseline design they started with.

A key benefit of generative design is that it lets JPL team rapidly iterate and improve designs. “As a design matures and additional performance or environmental data comes in, generative design lets engineers quickly create new designs,” says Willis. Most design teams typically take two to four months to revise a design, he points out. Working with generative design, that process can take be compressed into two to four weeks.

The interplanetary lander is tentatively designed to explore the moons of Jupiter and Saturn where it might find life or signs of it.

“The flexibility and speed to update existing problem definitions rather than starting from scratch, combined with the ability to specify manufacturing constraints, make generative design a real paradigm shift for people designing these kinds of structures,” Willis says. 

Creating a lander that can withstand the rigorous conditions of space across extreme distances is a massive challenge, and JPL’s designers are investigating and using every new technology at their disposal. For now, the application of generative design is still officially considered a developmental research project within JPL.

But taking the seemingly impossible and making it possible is JPL’s specialty. Just as the computational power of mainframe computers helped the space program succeed in the 1960s, technologies such as generative design are creating new possibilities in space exploration, opening the doors for us to go further and learn more about our place in the universe.

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish