Attempto! 02/2023: Rubble as a Resource

The idea came at dinner with a recycling entrepreneur in Germany. Between tomato salad and onion roast beef, he told the four friends about the truckloads of construction waste that are dumped at his facilities. “Pebbles, ceramics, bricks and cement are all mixed up. But a large part of rubble is actually recyclable.”

His friends quickly identified the business potential of solving this problem. During their studies, they experimented with cameras that could distinguish surface structures from fruit with the help of artificial intelligence. There was no reason that this method could not be applied for sorting construction waste. After raising capital, the four signed the startup contract in February 2022.

Optocycle’s office is based in the middle of Tübingen, on the Holzmarkt opposite the church and half-timbered houses. Inside there are desks with many screens, and a box with broken bricks and gravel on the floor. A camera hangs on a scaffold above the pile. The apparatus is a demonstration model. Optocycle’s actual facility in Lahnau, Hesse, films trucks entering a landfill.

Adults can easily tell the difference between a building and a truck, and whether a truck is loaded or not. Algorithms must learn this ability with the help of training data. Optocycle has a unique data set worldwide through its partnership with the recycling company in Lahnau. “Every day we get hundreds of images from the landfill and feed the AI,” says founder Lars Wolff, 31, who is the computer science expert in the team. 95 percent of the materials are already correctly recognized by the AI. The rest of the uncertainty will soon be gone. In a second step, the neural network must detect and compensate for the movement of the trucks so that clear images canbe created and evaluated. “The trucks pass quickly under the camera and distort the image,” says Wolff. His gaze follows the lines on his monitor full of letters, numbers and characters. He types something on his keyboard. A student intern at the University of Tübingen helps him to refine the code.

Meanwhile, cofounder Max-Frederick Gerken, 28, is at the camera testing desk talking to another recycling company based closer to Tübingen. There is no shortage of construction waste: 80 million tons of concrete, gravel, bricks and other debris are generated on construction sites in Germany every year. This makes construction waste by far the most important type of waste in Germany.

“The construction industry claims that construction waste is recycled, but this statement is misleading,” says Gerken. “The rubble is largely filled in road construction – it’s not really a circular economy.” Concrete and bricks are valuable building materials and could be recycled – if the trucks and their cargo could be divided into concrete, mainly concrete and mixed rubble. If this is done by human eye, too many mistakes are made. They rarely check the entire cargo and have bad and good days. Artificial intelligence can do this faster and more accurately.

Technology would be a contribution in the fight against climate change. “Eight percent of greenhouse gases come from the production of cement,” says Gerken. “If we solve the recycling problem of construction waste, we can actively contribute to the reduction of CO2 emissions.”

Together with the Cognitive Systems group at Tübingen University, led by Professor Andreas Zell, Gerken applied for more than half a million euros from Invest BW, an innovation funding program offered by the Baden-Württemberg Ministry of Economic Affairs. The computer scientists around Zell are developing a new method for the optical analysis of surfaces and materials with the help of artificial intelligence: so-called hyperspectral data cubes.

For the recordings, Varga places individual chunks from the crates on a plastic tray under his camera. The camera records both in the visible light wavelength spectrum between 350 and 750 nanometers and beyond up to 1,700 nanometers. “The spectrum beyond visible light is probably better suited to sorting construction waste,” he says.

Varga knows from experience. Professor Zell and his team are specialized in the hyperspectral analysis of food. They develop methods for the visual analysis of fungi on crops or the ripeness of fruits and vegetables and know how rich the information in the wave spectrum is beyond visible light.

This time it’s about rubble. Leon Varga and Hannah Frank hope to publish their results in a scientific journal. All findings will be publicly available – as scientists they are committed to open research. However, Optocycle can use the findings to develop an affordable camera for landfills. The camera in Varga’s laboratory costs several tens of thousands of euros and creates hyperspectral cubes in many wavelength ranges of the rubble.

After the experiments, Optocycle will know exactly which wavelengths are important for sorting construction waste. A specialized camera is much cheaper to develop. “Our camera should not cost more than a few hundred euros,” says Gerken. In a year and a half, he is confident that Optocycle will already sell the product.

For this last step of the founding phase, Gerken has tapped preseed financing in the amount of 250,000 euros from L-Bank and a co-investor, a construction company based in Stuttgart. The startup center at the University (see box) helped him to apply for the funding. “If the company is successful, the state can convert the loan into shares in the company – a funding model that was developed and successfully applied in Israel.

Gerken trusts that the construction of its start-up will proceed according to plan: one and a half years to market maturity, six years to the “Hidden Champion” and then something new. The fact that competitors could be faster does not bother him. “Before Mark Zuckerberg founded Facebook, he could have listed many reasons he might fail,” he says. “But he did not. Instead, he started.”

Text & Photos: Tilman Wörtz