I am peering through the glass window of a refrigerator-sized machine. The machine is named Endurance, if you go by the printing on its side, or Lucy, if you go by what Leila Madrone calls it. I’m watching some plastic get tortured.

It’s going through the equivalent of 100 years of life in a harsh desert climate: It’s been exposed to extreme heat and cold, and UV radiation. It’s been sandblasted. It’s been shaken around a whole lot. It suffers, because it needs to last 30 years without anyone having to fix it. Better for it to fail now, in the lab, than later, at a solar installation in some far-flung desert.

The building, in the former industrial sector of San Francisco’s Mission District, is older than it looks: it was used to make custom mining equipment during California’s silver rush in the early 1900s. Ideally, the plastic would last as long as this building. Maybe it will. If you have a problem with plastic, Madrone tells me while we peer through the glass, if you have a problem with the way that it sticks around in the environment, you just need to use it for what it’s good for — use it for something that’s supposed to last forever.

This plastic is part of the answer to a question that Madrone found herself asking back in 2008. By that point, she had built a lot of cool robots. Her thesis in electrical engineering at MIT was a set of motorized laser guides to help people play the theremin. After years of designing precision robots for biotech, she had fulfilled the life goal of her 7-year-old self and worked with NASA — leading an engineering team working on the Gigapan, a commercial version of the robot that lived on the Mars Rover and took panoramic shots of the places it visited. But Madrone still wondered: What was the most useful robot she could build?

Grist thanks its sponsors. Become one.

leilamadrone_headshot

Otherlab

Madrone settled on solar trackers — mechanical systems that move solar panels in order to help them follow the path of the sun. The concept isn’t new; solar-tech historian John Perlin found a description of a simple solar tracker dated 1699. (It was a slab that grapevines grew along the side of, that could be tilted throughout the day with a series of pegs and tracks to maximize sunlight.) Plants had come up with the concept even earlier.

But for someone with Madrone’s skills in automation and robotics, solar tracking looked like the right problem: It took a proven renewable technology and made it even better. Trackers in general could already boost the energy production of a solar panel by 20 percent or more. With the right breakthroughs, they could be the thing that tipped the balance and made solar the lowest-cost energy source out there.

Madrone read everything she could find about solar — in academic journals,  on the internet. She interviewed people working in the field. She began working for a solar startup called GreenVolts, which was developing equipment for concentrated solar: small, high-tech power plants whose modest footprints made it easier for them to locate near cities. GreenVolts’ technology was far out; the arrays looked sort of like a field of robotic flowers, with panels that branched out from a central column like petals. The panels, and the trunk, were actually a huge precision robot that could track and concentrate sunlight much more efficiently than old-school photovoltaic solar.

Grist thanks its sponsors. Become one.

It was really cool. It also came with a lot of bells and whistles — it was, as one business reporter put it, “bling-heavy.”

In 2008, the price of ordinary, non-high tech photovoltaic solar panels began to fall, sharply. It kept on falling. This was partly due to the recession, partly due to (successful) efforts by Chinese solar companies to seize the day and drive their competitors around the world out of business by selling photovoltaic panels at below cost. There was no bling that could outcompete that kind of bargain, especially when the bargain was a proven technology that everyone in the industry knew how to install and maintain.

GreenVolts was toast, though it would take a few more years to figure that out, and Madrone was disillusioned. She left the company and spent five months traveling through Europe, Asia, the Middle East, and Mexico. She visited islands where the only power source was expensive and dirty generators, and cities like Kathmandu, where the electrical grid was so non-functional that blackouts were a daily occurrence. The more that she saw, the more she became convinced that any solar technology that would have an impact needed to be so cheap and easy to manufacture that it could be deployed anywhere.

When she returned, Madrone began collaborating with an old friend at MIT, Saul Griffith. Griffith was the sort of inventor who came up with new devices and companies at the same clip that other people might, say, take out the trash, or do the nightly dishes (he’s best known for founding Instructables). At that time, he found himself increasingly preoccupied with climate change — and the problem of inventing tools to fight climate change that people actually used.

Among the projects he worked on: a road that is also a solar panel (turns out you expend more energy building it than you ever get back from using it). A wind-energy device called Makani Power, which is like the lovechild of a windmill and a kite (bought by Google X in 2012, where it has remained perpetually in the testing stage ever since). A cargo-hauling tricycle with electrical pedal assist (cool, but the battery alone costs $1,000, so not practical to bring to market). A website called WattzOn that catalogues tools to help people, cities, and utilities reduce their personal energy consumption (popular with a geeky subset of people, cities, and utilities interested in doing this, but otherwise under the radar).

Griffith had recently founded Otherlab, a research and development company that Madrone describes as part startup, part academic lab. Griffith and another Otherlab co-founder, James McBride, had also written a concept paper two years earlier, hypothesizing that mass-manufactured, plastic parts could dramatically bring down the cost of a solar tracker. Madrone moved into Otherlab and started building prototypes in a corner of the building.

Concentrated solar power (CSP) was still desperate for good trackers, so Madrone and Griffith started a new company — Sunfolding — and decided their niche would be designing a solar tracker that could work with any concentrated solar project. “We started with concentrated solar, thinking if we can do this, we can do anything,” says Madrone. CSP trackers were the hardest to build; they needed to be extremely precise while operating in harsh conditions.

What they came up with was a collection of small, inexpensive mirrors that tilted using pneumatic pressure. The whole setup had much less wind drag and fewer moving parts than a traditional steel-and-glass heliostat. In November 2012, the project won $2.6 million from the Advanced Research Projects Agency-Energy (ARPA-E), a government agency designed to fund risky but interesting energy technology, in the tradition of the Pentagon’s DARPA.

But the price of photovoltaic solar continued to fall, which made even believers in concentrated solar (which relies on an entirely different technology) balk at building new $3 billion plants. And the technology kept changing, making investors risk-shy: No one wanted to install a huge solar array, only to have to rip it out and start over a few years later because the technology didn’t work out. “There’s a lot of really neat innovations you can come up with related to energy,” says Madrone, “but they’re not necessarily what the industry needs or wants.”

What the industry needs and wants, Madrone and her colleagues hope, is huffing pneumatically a few feet away from where I’m sitting to interview Madrone, in an empty workshop on the third floor of Sunfolding HQ. It’s just around the block from Otherlab HQ, where the plastic is being tortured. This building, also old, was erected by a German-American pipe organ company that’s almost as old as the silver mining equipment company, but that proved much more long-lived. Schoenstein & Co., the pipe organ manufacturer, outgrew the space a few years ago and moved its operations to Benicia, Calif. Outside the door to the workshop is a huge room where, from 1928 until 2004, pipe organs were built. The list of each organ made on site, and its destination, still hangs from one wall.

Today, heavy-duty engineering texts like Uhlig’s Corrosion Handbook (third edition) are scattered around the upstairs workshop. Under normal circumstances, the ping pong-table-sized solar panel in the corner of the workshop would not be moving this fast, but right now it is rapidly tilting, fast-forwarding its way through a sequence of pre-programmed days. One of the things Madrone and Griffith found is that it’s easier to program a field of solar panels with pre-existing algorithms describing the sun’s movements than it is to fuss with a sensor that follows the path of the sun in real time — but would also be one more part that could break.

Standard-issue solar tracking devices for PV are designed, in Madrone’s words, “the way you would expect an industrial machine to look like.” They’re made of motors and gear boxes and bearings, which have a lot of surfaces that rub against each other and wear out. They have to be assembled by hand in a factory somewhere. The panels are moved by a torque tube that runs along the length of 20 or 40 solar panels, so that it can move all the panels at once when it rotates. This is efficient, and doesn’t use much energy, but has the unintended effect of forcing every large-scale solar installation that wants to use tracking to become a giant, flat rectangle — no matter how many nice, oddly slanted or shaped spots are nearby.

Sunfolding’s approach is different. The only hardware that moves Madrone and Griffith’s test panel is a set of chunky, accordion-pleated tubes, about the size of a liter bottle of soda. They are linked via a series of tubes to an air compressor off to the side, and as air moves in and out of the tubes, they expand and contract, and the panel tilts. The tubes are made out of black automotive plastic; as they expand and contract, the effect is part inchworm, part Muppet.

I ask Madrone if, compared to all the other robots she’s built, this one might not be just a little boring. “I’m glad that you see it’s a robot,” she says. “Because it runs autonomous systems. It runs presumably for decades without anyone telling it what to do. It responds to the sun. If there’s bad weather it does something to protect itself. For all intents and purposes, it is a robot.

“It’s not a fast-moving robot. It doesn’t move any faster than the sun moves. But a lot of the things you have to solve for this tracking system are really interesting problems. There are things we are doing with our controls that no one else can do.

“It’s really easy to build something that works for a day,” Madrone adds. “It’s a whole other piece of work to build something that can go for a week. Every time you increase that without someone coming in and poking it and replacing things, it becomes harder and harder.”

It has not escaped Madrone’s attention that when she graduated, many of her fellow engineering students at MIT worked on hardware projects. Now, more and more of them work on software. Even with the much-hyped “internet of things,” and the tsunami of venture capital flowing through the Bay Area, the field for hardware projects is small. Companies that can invent new hardware and actually bring it all the way to market are rare. Venture capitalists may come in at the late stages of a product’s development, but they aren’t going to fund research and development for hardware when software, immaterial and scalable, is so much more tempting.

Only five years passed between Madrone’s decision to build a cheap way to move solar panels around and her actually releasing a product to market. That’s an eternity in internet years, but a pretty short one for hardware that is supposed to make it to 2045 before it needs repairing.

Getting to this point took more than building a better mousetrap. It took learning the ins and outs of an industry that was changing fast, yet skittish about changing any further. It meant leaving cool ideas by the wayside if they interfered with cost and reliability. It meant building something most individuals will never see in action (since building codes have a hard enough time permitting rooftop solar that stays perfectly still).   It meant signing on to a funding mechanism that involved spending a long, eight-hour day every three months talking technology and strategy with a government agency, ARPA-E, that was both a champion and an enforcer. (Madrone sees this as a good thing: “A lot of startups don’t have that experience where they have someone scrutinizing them and making sure they are doing everything technically correctly.”)

The tech industry, as a whole, is prone to big visions and big ideas — with good reason, since big ideas are the language of venture capital. That’s not the language that Madrone speaks.

“We have to get away from this hero mentality,” she says, as we leave the 100-year-old pipe organ factory and step out into the sunshine. “We don’t need someone to create a magical box that means we can do whatever we want all the time and use all the energy we want. We need to get really smart about our energy use and really smart about how we create energy.

“The only way we can do that is by creating an ecosystem where there are a lot of ideas working together. And we need to start valuing that kind of community of ideas instead of the one hero that is going to save us all. ”