Making recycling more effective with robots
Reduce, reuse, recycle – we’re all familiar with the mantra, and most of us dutifully sort our waste. But what happens after that? The fact is that a lot of what we “recycle” still ends up in an incinerator.
This happens because a certain minimum level of purity – as defined by having a relatively clean, consistent type of material – is needed to recycle items like glass, plastic, and paper. And the only way to guarantee this purity is by sorting waste, which is a time consuming, labor intensive, repetitive and sometimes dangerous job for humans (and also makes it difficult to recruit people to do it). Sorting is also needed to recover valuable materials, like copper, from waste streams for reuse.
Robots to the rescue! ZenRobotics uses artificial intelligence (AI) methods to improve the quality of waste sorting. Their unique AI algorithms mean that robots can sort waste on average as well as humans do – just faster and more safely, working tirelessly around the clock. Reaktor worked with ZenRobotics to help them roll out their waste-sorting robots faster, while also helping the company to work in more agile ways.
- ZenRobotics combines data from multiple sensors to create an accurate real-time analysis of the waste stream. Based on this analysis, the robots make autonomous decisions about which objects to pick and how
- ZenRobotics robots use machine learning methods to handle a wide range of waste streams – from construction and demolition waste to plastic, paper, and metal sorting
- Reaktor worked closely with ZenRobotics to further develop configuration management for the robots, allowing faster future growth and product roll out
- Reaktor introduced the customer to new ways of working to help make their processes more agile and efficient
The dirt about recycling
In order for materials like plastic, glass, and paper to be recycled or properly disposed of, they need to be sorted into relatively homogenous groups of material called “fractions”. To take just a few examples, there are seven different categories of plastic (some of which can’t be recycled at all), paper can’t be recycled if it’s lined with plastic, painted and unpainted wood have to be treated differently, and even a small amount of food waste can render something unfit for recycling. If the level of purity in a given fraction is not high enough, the material can’t be recycled and will instead be incinerated or sent to a landfill. In addition to recycling, material recovery is also important – separating valuable metals like copper from other waste streams, for example. In order to create a circular economy where materials are actually reused and recycled multiple times, efficient sorting is a must.
Meet the pickers
ZenRobotics created their AI-powered robots, called “Heavy Picker” and “Fast Picker”, to be able to adapt to a wide range of sorting needs. These sorting robots can separate multiple waste fractions with the help of various sensors, heavy-duty robot arms, and artificial intelligence. Robots also have the advantage of never getting tired and are able to perform repetitive tasks that often require moving heavy objects.
Especially in construction and demolition, every waste stream and processing center is different, which means Heavy Picker robots are trained on the premises using machine learning methods to ensure they can perform as needed. For uniform waste streams like plastic bottle recycling, this step is not needed as Fast Picker can work by using transfer learning. The end result in either case is an unmanned sorting process that makes waste sorting safer and more cost effective.
Working closely together
Reaktor was brought on board to work with configuration management, the process that allows for faster and more efficient roll out of the robots on a given site. We worked on the customers’ premises to make sure we could help as needed on a day-to-day basis. We also helped with “dead code”, simplifying maintenance by removing code that no longer serves a purpose.
Because we were working onsite and in close collaboration, we had the team from ZenRobotics interview our consultants to find the best fit for them – we never expect our clients to just assume that we know what we’re doing, we think that by testing us it helps them to trust us.
Improved ways of working
We introduced new ways of working to the customer to help them be more agile. These included practical tools like kanban boards, dailies, and retrospectives. At Reaktor, we have an extensive toolbox of methodologies that we use from a wide variety of different cases, which means we can choose what works best for any given customer.
In addition, we helped ZenRobotics with their recruitment pipeline. To help minimize turnover, we focused on the onboarding process and suggested that they expand their documentation, as that would help new hires to quickly get the knowledge needed to work efficiently and understand what is going on. Checklists were also introduced – a simple but effective tool that has been applied with great effect in fields ranging from coding to surgery to airplane maintenance.
How we helped ZenRobotics to work in more agile ways
Use of Kanban boards to better manage workflows and prioritize tasks
Holding short daily meetings to ensure everyone knows the project status and can share progress
Use of checklists to minimize mistakes
Holding retrospectives to discuss what went right with a given project – and what could be done better next time
Implementing career growth discussions with peers to ensure meaningful feedback and career development opportunities
Read more about Reaktor career growth discussions: growthdiscussions.reaktor.com