1660342896083 2photo3

Autodesk Futurist Peers Into the Technology Crystal Ball

Dec. 10, 2014

Diving deeper into the conversation started by Autodesk CTO Jeff Kowalski and CEO Carl Bass, among others, Jordan Brandt, Autodesk Technology Futurist discussed trends that industry is likely to see at Autodesk University, held December 1-4 in Las Vegas.

As mentioned in a previous blog, sometimes it helps to step outside of traditional control and automation approaches and see what engineers are doing in a world that is parallel yet does converge in some areas. Diving deeper into the conversation started by Autodesk CTO Jeff Kowalski and CEO Carl Bass, among others, Jordan Brandt, Autodesk Technology Futurist discussed trends that industry is likely to see at Autodesk University, held December 1-4 in Las Vegas.

Jordan Brandt, Autodesk Technology Futurist discussed trends that industry is likely to see at Autodesk University, held December 1-4 in Las Vegas.

Brandt is a huge proponent of 3D printing, and not only for consumers. "The real benefits will go to industrial manufacturing operations," said Brandt. "For instance, companies are seeing big increases in productivity by printing injection molds instead of the parts themselves. As such, a helpful program at Autodesk is Spark, an open software platform that makes it easier for manufacturers, software developers, scientists and designers to connect any file to any 3D printer, similar to how Adobe's Postscript software standardized desktop publishing."According to Brant, one futuristic trend is that of computational materials, or substances designed by a computer to have the superior properties required by engineers. Here, software figures out the materials' recipe based on rules entered into the computer by engineers. "The same approach of letting the machine figure things out can be applied to the design of chairs, machines, buildings and bridges," he said. "In these scenarios, you can design anything without even knowing what it's supposed to look like."

Another hot topic is that of  generative design, which is about getting machines to be more intelligent about helping engineers design and manufacture products, continued Brandt. "Sketching is inherent to all designers and engineers. It involves what we call 'explicit design,' meaning you have the idea in your head and you draw it out. You draw as quickly as you can and then iterate on the design. Iteration is core to all design, whether you're designing buildings, cars, consumer products, industrial machinery, or whatnot. An app of  interest here is Shapeshifter, which exists purely in the browser. It allows individuals who are not professional designers or engineers to do iterations of various designs based on templates with sliders that govern size, shape, the pattern and tessellation. The technology ensures that the object is 3D printable, meaning it builds manufacturing constraints into the intelligence of the system itself. And a more sophisticated tool like Dynamo is a visual scripting language that lets you algorithmically generate forms and quickly perform geometrical iterations. So rather than drawing an object directly, you're teaching the machine how to draw it for you."

Also of interest is BIM 360 Glue, a platform that lets stakeholders collaborate in 3D in real-time, which boosts iterations exponentially, said Brandt. "Going back to the idea of sketching, designers have an inherent desire to iterate to solve their problems. They just needed a tool set that made it easier to perform iterations. If every time you wanted to brush your teeth, you had to do 20 minutes of preparation and create the toothpaste, you probably wouldn't brush your teeth as often. Similarly, it was too hard for designers to collaborate in the past."

According to Brandt, the idea of iteration has even leaked into new business models such as practiced by Google. "Companies are trying a business model and then modifying it over and over. And that's good, because we need to borrow from other industries to advance the next generation of our tools. In designing a product like an airplane, objectives or design goals might include increase fuel efficiency and the stiffness of the wing, while decreasing weight. Machines should be able to augment our decision making such that they map multiple solutions and how well they performed. A person cannot digest this amount of information."

For example, Moon Express is a company sending a private craft to the moon in 2015. The company is quickly and algorithmically generating parts for the craft that would take a year for a person to draw. "This is where the cloud comes in because such designs would be impossible to create on a desktop machine. Software on the cloud lets you generate three variations per second, and go through generations and generations of variations to develop an optimal design," added Brandt.

An interesting  challenge is that iteration is not intelligent, claimed Brandt. "In fact, it's  much like evolution, taking a billion shots in the dark and hoping that one fits the criteria. However, evolution proved that it works. For example, humans, the product of a bunch of random mutations, are doing okay—hopefully they won't screw things up. That's why we build technologies to investigate the process of synthetic evolution. Consider Project Cyborg, a system to model and simulate molecular behavior, which would give you a blueprint to 3D print biological tissues."

Newer image recognition technologies are even using algorithms that teach machines how to recognize images. "You can fed the system millions of images, and let it develop its own system of learning, its own series of neural networks, to filter out and find what is significant in the image to recognize what it represents. For example, a paper shredder is a box that might sit in the corner of your office or house," said Brandt. "Image recognition algorithms can now identify makes and models of paper shredders better than humans can. In other words, the machine is generating its own code to recognize a given image. That means that if the original developer of such an algorithm looks at their own code 10 weeks or two years later, they're probably not going recognize it anymore, because the code has modified itself based on how well it's performing. Other industries have been doing this for a while; the question is how do we apply this to design? "

Consider all the data organizations are now privy to. "Boeing has designed a bracket countless times and the same is true when it comes to Ford designing a gear," continued Brandt. "What if we could collect all this information and also see how design iterations were made over time. If we could learn something from each change, we are converging on a solution. Products such as Fusion 360 potentially give users access to enough data to make meaningful sense of design iterations. That said, even given the advances of automation, robotics, and design automation, we still have to realize that a robot isn't a craftsman and an algorithm is not a designer, so engineers will be more necessary than ever before. Futuristic technology doesn't involve  a decision between automation or humans, it augments humans."

Sponsored Recommendations

High Sensitivity Accelerometers to Monitor Traffic and Railroad Vibration for Semiconductor Manufacturing

This paper examines highly sensitive piezoelectric sensors for precise vibration measurement which is critical in semiconductor production to prevent quality and yield issues....

Simulation for Automation Guide

How digital twin solutions are expanding the capabilities of plant engineers.

Enhancing HMI Security and Accessibility with Cloud VPN Solutions

Enhance HMI security and remote access with Beijer’s cloud VPN solution. Enjoy advanced encryption, easy setup, and secure access via laptops, smartphones, or tablets. Cut costs...

Motor Encoders: What They Are and How They Work

Motor encoders are rotary encoders adapted to provide information about an electric motor shaft's speed and/or position. Like rotary encoders, motor encoders are most commonly...