Generative design, a long way from home.

Opportunity is dead.

After being expected to survive for around 90 days, the Opportunity rover transmitted its last signal an incredible 14 years after it touched down on the Martian surface (a few weeks after its partner and identical twin, Spirit).

Last contact with the iconic, 180kg vehicle was on 10 June 2018, before a global-level dust storm cut off its energy source (solar power) and telecommunications. On 13 February, John Callas, Manager of the Mars Exploration Rover (MER) project at NASA’s Jet Propulsion Laboratory (JPL) said in a press release that after over 1000 attempts to re-establish contact, it wasn’t feasible to continue.

Opportunity now rests in state, having given us more scientific insight and inspirational images than we could hope for – not to mention the record for driving the farthest distance in a single day on another planet.

Now, the quest for signs of life in the Solar System continues, and because Mars is only in our backyard in astronomical terms, we have to look further afield, particularly to the moons of the gas giants Jupiter and Saturn. The latter’s largest moon, Titan, for example, has a dense atmosphere that was suspected as far back as 1903 and confirmed in 1944.

But travel to the outer Solar System will call for more advanced technology than those we’ve used for space travel for the last 50 years. Where Mars is 55 million kms away from the Earth at its closest – a trip of about five months – Jupiter lies between 365 and 600 million kms away and Saturn 1.2 billion kms at its closest to Earth.

For a trip of over a billion kilometres,  we’ll need sweeping changes to suit vastly different mission constraints, and a 2.5m wide, 1m-tall prototype lander created by a collaboration between JPL and Autodesk might be part of the answer.

Have algorithm, will travel

This lander concept might be the most complicated project ever built using generative design. When Autodesk approached JPL to suggest working together on the next generation of extraplanetary landers they were open to the idea, but cautioned that incremental performance gains wouldn’t cut it. They needed a paradigm shift.

It was found using the more top-down view characteristic of generative design. The ‘old’ way of designing space-borne landers was to identify components and where they’re located, then manually fashion a structure around the resulting layout.

When provided with the parameters around the components and their position in the craft, generative design came up with the strongest, highest-performing structural framework possible.

One of the most critical gains of many engineering projects is reduced mass, and as you can imagine, that was a very important selling point for JPL. In fact it was when Autodesk told JPL engineers they could bring mass down by up to 30% over the existing baseline design that the project really got their attention.

And nowhere is weight more critical than in space exploration, where every pound is agonised over, deployed and counteracted. Every extra gram at lift-off weight means more fuel necessary to get the whole thing off the ground, which adds to the weight, which means more fuel in turn, and so on. In fact, every two pounds the Autodesk team saved in the lander reduces the weight of the lift-off vehicle by up to a ton.

And of course, representations of the final generative design can happen much faster than traditional engineering designs. Where traditional methods can take between two to four months for a project of this magnitude and complexity, the Autodesk team compressed the timelines down to just a few weeks.

That doesn’t only make the project faster, it’s more responsive to change. As we all know, the human race has a pretty impressive fleet of sensors and measurement tools throughout the Solar System today, and they’re always reporting back with new information about the conditions the lander has to operate in.

If data from another mission makes us realise a given material might be vulnerable to corrosion or radiation it might throw a long and very expensive research path out the window. Comparatively fast updates to the relevant manufacturing constraints (like materials inputs) in generative design could get you back on track much faster.

Going live

Space exploration is a business with its own economic peculiarities, many of which don’t apply to the for-profit manufactured goods sector. In other industries it can be considered a good thing to burn through resources to get to a minimum viable product as fast as possible, using the experience and prototype to build out to your production model.

But for obvious reasons, spaceflight is a one-shot proposition. If your build architecture fails in orbit around Saturn, the nearest technician with a wrench ready to make the necessary tweak is several years (and probably a few more billion dollars) away. Even if it’s a software or drive systems issue it’ll take over an hour for the instructions to reach the craft by radio – a further hour for its confirmation of success of failure to reach us back home, and in the critical minutes of spacecraft landings, anything can happen.

All of which means your first and final build has to be perfect straight out of the gate. And with taxpayers and their political representatives only too ready to pounce at any sign of wasted funds, the pressure is even higher.

To make JPL’s position even more fraught, it finds itself today in what’s very much a business climate complete with competitors like SpaceX, so to not adopt new technology when it’s available is a sure-fire way to be left behind. Like so many other industries have discovered, generative design is a quantum leap forward.

So far, the lander is still a research and development project for JPL, far from sign-off and production in the space program. But new approaches to designing and making things have taken us from the first extra terrestrial satellite 60 years ago to robots walking or rolling across other planets – who knows where generative design might fit in?

www.autodesk.com.au