Compology Goes Dumpster Diving for Data
Modernizing Authentication — What It Takes to Transform Secure Access
An upstart California company introduces sensors and software to make waste collection far more efficient.
Few businesses are as inherently inefficient as waste collection. For a typical firm that collects trash from multiple sites, a fleet of trucks canvases routes and empties commercial Dumpsters whether they contain only a few items or they're overflowing with waste. "Companies spend a lot of labor hours, gasoline and other resources running the business in a highly inefficient manner," said Matt Duncan, software lead for Compology, an upstart San Francisco firm that has developed systems for automating waste collection.
The company, which collaborates with major players in the waste-collection industry, has developed a cloud-based dashboard and connected sensor system that provides data about bins and Dumpsters, thus allowing collection firms to use dynamic routing. These containers are as small as two cubic yards and up to 40 cubic yards at major construction sites. All are retrofitted with a sensor that snaps still 2D photos of the inside of the bin every two hours. The system uses cellular connectivity to transmit the data to a centralized dashboard, where software translates the data points into actual information about whether the container needs to be emptied.
"The result is that waste collection doesn't have to abide by a set schedule and follow the same route every day or every week. We're able to route collection trucks based on actual demand and the Dumpsters that need to be serviced," Duncan explained.
If a company wants to empty bins when they are 75 percent or 90 percent full, it can do so. It can also create different thresholds for different businesses or sites. It's possible to manage an entire network of Dumpsters and change truck routing that morning or in near real time, he said.
Compology relies on a machine learning platform from software vendor Dato to drive the initiative and provide the data modeling capabilities. Overall, the system automatically classifies between 40 percent and 50 percent of the images and achieves greater than 96 percent accuracy. Duncan said that the company is continuing to work on the underlying algorithms and machine learning models to drive further improvements.
The goal, he added, is to scale the technology out to tens of thousands of trucks scattered across the country and beyond. "There is an opportunity to not only manage a fleet of trucks far more effectively and efficiently but also significantly cut fuel costs and operating expenses," he said.
Compology is not the only company looking to tap into the Internet of things technology in order to talk trash. For instance, Needham, Mass.-based Bigbelly has installed more than 30,000 connected solar trash and recycling receptacles around the United States. Waste haulers can view the capacity of a container at the click of a button on a computing device. Another firm, Sutera, in Greenville, S.C., operates connected semi-underground waste bins. A Finish firm, Enevo, is now marketing a sonar sensor-based-system that allows waste collection firms to optimize operations using open APIs.
"We are entering a new era of waste collection. Technology is making the process more efficient and less expensive," Duncan said.