He noted then that Dreamcatcher was still a research project, but that there would be “some of this generative technology coming out in the next months in Netfabb.” Later during the show, I talked with Duann Scott, Manager, Strategy and Business Development for Autodesk’s Digital Manufacturing Group, and asked if he might have more he could touch on regarding Project Dreamcatcher. While in the days between McManus’ keynote and my conversation with Scott I had heard of other journalists being rebuffed in efforts to learn more about Dreamcatcher, to my happy surprise Scott was able to share more information as Autodesk began to ease toward announcing more.
“Can the computer do the busy work?” McManus asked. “Let’s not only think about additive; think about whole spectrum of how digital manufacturing works.”
That Autodesk was beginning to open up about the platform was promising, following a few years of experimental projects and trend toward generative design. Several high-profile projects have incorporated generative design as Autodesk has, for example, worked with Airbus on developing an aircraft dividing wall in 2015, as well as recently creating a proof-of-concept research design for a lighter weight airplane seat frame. It came as little surprise, then, when last week Autodesk announced a big addition to the upcoming Netfabb 2018.
“Generative design offers cloud options to explore, through which users can pick the most important properties,” Scott told me. “What materials, what processes are available to me? From there, you get a guide the software generates for you, and you can go with that or use as just a guide for further design work. There is usage you could not have conceived of in your waking life.”
Last year’s release of Netfabb 2017 included an increase in design optimization, and the 2018 version is set to bound beyond that. I had the opportunity to gain more information about the new offerings through a chat with Greg Fallon, Vice President of Simulation at Autodesk, who has expressed his enthusiasm for the power of generative design as the announcement made its way out into the world.Fallon, for his part, is just rounding out his second year working with Autodesk, as the work with advanced capabilities had drawn him to the company in the first place. He runs the Simulation group, which makes all the products Autodesk delivers that have to do with simulation, as well as overseeing the commercialization of Project Dreamcatcher — which is now Autodesk Generative Design.
“You know we’ve been working on this concept of generative design for quite a while; it’s been about four years. It started as a research project, as Project Dreamcatcher. Now we’re ready to start releasing products in the coming months,” Fallon told me. “This is a big thing for Autodesk, for the industry, for our customers. Generative design is not just a feature we’ll be releasing in Netfabb; we think it will really change the paradigm.”
He was quick to key in on not only what generative design is, but what it isn’t. He made this point as well in the announcement he wrote covering the new features, focusing on how all-encompassing a solution this technique is.
We discussed some of these capabilities, and manufacturability became a throughline in the conversation. Generative design is set to “encompass the entire process, from ideation to inspection,” he explained, and its capabilities heighten the smarts with which a system can operate, offering benefits to not only its operator but to an entire process, or company.
“Autodesk Generative Design is not just topology or lattice optimization alone – it’s a massive step beyond that,” he wrote. “While optimization focuses on refining a known solution without any notion of manufacturability, generative design helps the engineer explore a whole cadre of functional and manufacturing design options. With Autodesk Generative Design, a designer or engineer can not only discover a new solution, they can then bring it to life using additive manufacturing tools.”
The travel analogy clearly illustrates the differences between the two design approaches; while the best way to travel from New York to San Francisco via ship would include passage through the Panama Canal as indicated by a topology optimization-like method, rather than going around an entire continent, that precludes the notion that perhaps going via ship at all itself is not the best way to go about the method. Through looking to generative design, the trip would also be considered via other means of transportation, as going by foot, bicycle, car, bus, train, and airplane could all be considered as well. While not all the considered routes would be winners — probably not the best plan to hike it, or go by horseback — by considering all manner of different available options, it could just turn out that flying across the country would be the most expedient way of getting there. This does, of course, oversimplify, as in this example it may be a bit obvious what the best means would be, but taking this example and running with it, expanding it to a whole world of design options for myriad applications, can prove a better way of wrapping one’s mind around what’s going on here.
“People talk a lot about optimization in 3D printing to minimize material, through lightweighting, through latticing, removing material — typically topology optimization. That is very different from what we’re doing. We use topology optimization in generative design but that’s not all it is,” he told me.
“If you think about topology optimization and how it affects design, you can use a somewhat simple comparison: it’s like taking a trip from New York to San Francisco. Topology optimization will have the fastest route by ship, so you can avoid wasting time and fuel, but it’s not going to give you other alternatives from New York to San Francisco. You have to kind of say what you’re looking for, not only in how it works but how it looks. With generative design you just say, I want to go from New York to San Francisco, and it will think up other optimized routes and options that you might never have considered.”
Autodesk has chosen to start their roll-out of generative design commercialization with a focus on additive manufacturing — itself a new way of thinking about production.
He continued, noting that working with generative design effectively offers engineers “an assistant to help with the project” that will help them to become more creative, working with the requirements and constraints of design, and to connect that engineer with other parts of the company. Working through the complete process, from machine code to scheduling and tracking of a job, allows generative design to become completely immersed in the end-to-end process — through which it will intelligently learn about the design needs of a particular enterprise.
“We’re choosing to focus first on additive manufacturing for a couple of reasons,” Fallon told me. “This is a new manufacturing method, and companies in general who want to do additive already need to redesign their products and reexamine their processes. There’s more flexibility in additive, more flexibility of geometries and of materials to consider. We’re releasing generative design to users within the Netfabb suite this year, and will eventually take it to other offerings like Fusion 360.”
As we drew back to the idea of machine learning and algorithms, I asked just how smart this process might be.
“We’re just getting into this, and the possibilities are endless in how smart this can be,” Fallon answered. “I see generative design eventually being very personal to the engineer using it. It should know quite a bit about the organization and the people using it. At a rudimentary level, it will learn about the machines, the printers, and the materials you have on hand, and the factory schedule, so it will also understand the organization.”
He pointed to work that Stanley Black & Decker has done with generative design as an example, as the company recently redesigned a hydraulic crimper.
The crimper is a handheld tool that requires a lot of force to join wires, Fallon explained. The tool at the end of the overall device was redesigned using alternatives created via generative design that the human designers hadn’t thought of previously. This is allowing for “totally new ideas” emerging.
“In the Black & Decker case, Dreamcatcher will understand what the look and feel of Black & Decker is, and will ultimately design products that look like these,” Fallon explained of the learning capabilities possible with generative design over time. “This can get very abstract — but it’s very concrete. In the short run, algorithms are mostly focused in physics, are optimized around physics. I see this getting very personalized and very smart.”
As availability of generative design expands beyond a few major companies like Stanley Black & Decker, Airbus, and Under Armour, Autodesk foresees some major market possibilities. One area recently touched on is in construction, as sites can be digitized, as we see “an extension in terms of the convergence of construction site with manufacturing in general,” he noted.
“Generative design fits squarely in all those different opportunities, and this technology that will underpin Autodesk future growth, it will really change the way our customers design and think of products,” Fallon said. “It’s important for me to point out how different this is from other technologies on the market.”
Speaking of the future growth of Autodesk, I asked as well whether there would be any changes in store for the company’s upcoming releases or strategies in process following the recent announcement of a new CEO. The conversation turned a bit more party-line at that point, as Fallon said, fairly, that he “didn’t want to put words in Andrew’s mouth.” He could say, though, that the “goal is to live up to commitments as-is and move forward.” It would have been more surprising had big changes been on the horizon with the new executive leadership, as new CEO Andrew Anagnost has been with Autodesk for some time; we’re assured that it’s business as usual.Turning back to generative design as we closed out our chat, I asked Fallon one of my favorite questions for those behind big moves in technology: What’s exciting to you?
“What’s exciting to me is the pace of innovation and human creativity in general,” he enthused. “One thing I think about with generative design and Project Dreamcatcher is it allows engineers to do what they’re trained to do, which is make things. So much time is tangled up in documentation, in mundane administrative effort. Dreamcatcher gives the ability to take care of that, to really extend creativity. It’s helping people see opportunities they may not have already seen, to take care of details that take weeks or months behind the scenes and require administrative skills.”
The more artificial intelligence learns about operations, and the more experience generative design ha with a given endeavor, the smarter it gets. By putting more on the automation, the humans ultimately behind the work can get busy without the busy work.
We’ll be following up with Autodesk as Netfabb 2018 rolls out later in the year, and as generative design begins to take some of the mystery out of machine learning. Share your thoughts in the Autodesk forum at 3DPB.com.