It’s Alive! How Generative Design Marries Nature With Technology to Bring Objects to Life

by Jeff Kowalski
- Feb 2 2015 - 6 min read
generative design
Micke Tong

All of the amazing things ever imagined, designed, and created—buildings and bridges, cars and machines, and products and devices—have one thing in common: They’re all dead.

They’re the result of an inert process, and they lack the characteristics of life: They can’t sense what’s happening around them, respond to external stimuli, or collaborate with each other to get things done. But that is about to change.

Today, people take a techno-centric, brute-force approach to design. Emerging to take its place is a perspective that looks at design more like a living system.

Nature fundamentally designs by taking the best existing solution to a problem and iterating. But unlike nature, people often start their design projects from scratch. That approach leads to retreading the same old pathways and coming up with only incrementally better results, when what we’re really looking for are revolutionary outcomes.

shell as design in nature

Next-Level Computer Intelligence

For designers to experience similar progress to nature—which only moves forward—they need to consider the huge corpus of potentially relevant ideas and designs that already exist.

What if, like nature, you could access every design, schematic, asset, and idea that had ever been created to help solve a design challenge? Machine-learning algorithms in computers can now detect patterns inherent in millions of 3D models and generate taxonomies without direction or intervention by humans.

The computer learns what all of the elements and components are, classifies them, identifies how they relate to each other—such as how a gear collaborates with other gears, axles, and bearings—and what they do. It can then serve up dozens of different design options for a specific dimension of a component and provide them as elements for your next design.

For example, what if you need to design a device responsible for transmitting torque from one point to another? The computer responds with a set of potential design solutions, such as gears, linkages, and pulleys. If gears are the most appropriate solution to your torque problem, the computer will present various gear configurations.

With the computer working in a lifelike manner to solve problems, you can focus on the intention of your design—what you’re trying to achieve—while the computer uses everything it knows to provide you with the best solution.

generative design gears

Generative Design = Evolution

Nature explores all possible solutions that optimize performance in a given environment. For humans to see their designs evolve, they have to stop thinking of computers as merely drawing tools and start thinking about them as portals to greater exploration.

Generative design mimics nature’s approach to design by starting with your goals and then exploring all of the best possible permutations of a solution through successive generations, until the best one is found.

Until recently, generative design was locked up in the realm of theoretical computer science because the computation took too long. But because the cloud gives you access to all the computing power that this approach demands, you’re now able to evolve millions of individual options in parallel. That essentially accelerates evolution—and now you can get to the best result in the same time it once took us to get to the first result.

By identifying a specific parameter to optimize for—like weight, surface area, or stiffness—the computer will return the most optimal geometries. The designer can then choose the best of these options based on, say, aesthetics, and the computer will continue to iterate on that design until you ask it to stop. Eventually, you end up with this highly optimized result, which is not only beautiful but lighter and more efficient than its traditionally designed predecessors.

generative design for sculptural joint
ARUP’s sculptural joint. Courtesy ARUP.

This summer, engineering firm Arup used WithinLab’s generative design tools to create sculptural joints for tensile structures. As the forces within each joint are unique, each computed design is optimized and correspondingly unique.

Meanwhile, MoonExpress at NASA Ames Research Center is working on a lightweight bracket that attaches rocket thrusters to their moon lunar lander. Instead of reducing weight by removing material from a given geometry, they’re exploring ways to computationally grow the structure.

Community of Things

The current response to our changing world is planned obsolescence: When the world changes and our product doesn’t, we just throw out the old and bring in a new one. But what if design could live past the point of creation? If so, the things we design and make in the future must do three things: sense, respond, and collaborate.

The most rudimentary form of living design is an object that can simply sense something about their surroundings and collect that data. Things like smart meters, personal fitness devices, and instrumented industrial components are made possible by the proliferation of super-cheap, ubiquitously available sensors.

But collecting data—even big data—is not enough. We need our objects and environments to “respond”—to take some kind of action based on their sensory input. An example is a carbon-fiber spoiler on the back of a race car that raises up when it senses moisture, increasing downforce and improving traction on a slippery track.

Programmable carbon-fiber spoiler
Programmable carbon-fiber spoiler

People call this the Internet of Things, and that’s a problem, because it isn’t about things, and it’s not even about the Internet. It should be about what these technologies enable for us in terms of experience and value. The most enlightened things we design will be those that are actually capable of collaborating with each other to create new experiences for us.

When an entire city is filled with things and systems that can sense, respond, and collaborate with each other, it comes alive. Such a city could naturally understand the traffic and use patterns within its borders, predict changes, and decide on its own to redesign its infrastructure.

Since the term Internet of Things implies a false hope of an emergent experience, what we really need is a Community of Things: a collection of individual entities designed to actively and purposefully work together.

Looking ahead, the twin engines of technology and nature are becoming increasingly intertwined in a mutually beneficial dynamic.

Consider how we have traditionally dealt with the problem of nuclear remediation: When we go to clean up a disaster site, we either encase it in concrete, move it and bury it somewhere else, or just isolate the whole area and let it slowly burn for 10,000 years.

Nature has a head start on a different approach. Case in point: There’s a fungus found at Chernobyl that has evolved to thrive in radioactive environments. One Autodesk researcher is accelerating nature by re-engineering this radiotrophic fungus to capture and convert radioactive particles so they can be cleaned up safely.

Lawrence Livermore labs diagram
Lawrence Livermore National Laboratory materials diagram.

Then there are the researchers at Lawrence Livermore National Laboratory, who are architecting entirely new materials. In the history of fabrication, we’ve been restricted to that diagonal line of materials on the graph that nature has provided and what we’ve been able to recombine. Now we’re able to go off-axis and make things with surprising properties—things that expand when you stretch them or shrink when you heat them.

This is just the beginning. There are whole territories ahead where we can take nature’s strengths and extend them with technology’s power. By doing so, we can create things that transcend the limits of both and truly bring life to design.

More Like This


You’re in.

Get smart on the future of making things.

Subscribe to our newsletter.