Every time you throw waste into a designated bin, that waste eventually ends up somewhere and has to be treated in some way in order to not be hazardous to the environment. Over the years, technology has helped cities transform the process into a highly smart operation management activity. An IoT, ML-enabled platform for waste management adds a dimension of agile, real-time mapping and tracking that can improve waste management outcomes.
For major cities like Hong Kong, Seattle, Chicago and Seoul that accommodate large populations, dealing with waste disposal can be a challenge. However with the rise of smart cities, government administrations are able to provide a massive push to incorporate technology into every aspect of their city, including waste disposal.
Today, civic waste management in any smart city is an interplay of on-field devices, or sensors, all networked together to generate millions of data points; data thus obtained is then ingested into a cloud platform and fed through complex analytical frameworks to analyze and then to derive sensible, actionable inferences to better serve the citizens of that city. This creates a more efficient method as human error can be factored out. In fact, the whole process is automated with almost zero human interference.
We all know waste can be placed under a few broad categories; paper waste, plastic waste, food product waste, water soluble waste, water insoluble waste, animal waste, sanitary waste, domestic waste, industrial waste and so on. Some of them are biodegradable and some aren’t; some could even be radioactive waste, which is highly toxic and potentially hazardous.
Some solutions proposed solution to this involves the application of alerts where citizens can submit photographs of waste that is lying on the side of the road or overflowing from a collection bin to a command center via a cross-platform-compatible mobile app. The picture received by the command center can then be analyzed using an image point and vector framework analyzer to ascertain both the approximate quantities as well as the possible categories of the different waste materials captured in the image. The process flow requires no human intervention and it uses a smart algorithm to match with past and existing data. Nearly 90 percent accuracy has been achieved over time, specific to this activity.
The sensors attached to the roadside waste bins track waste collection inside the bin and alert the waste collection trucks automatically using an IoT-integrated and networked system, but waste that is thrown directly on the road escapes the digital visual eyes of the sensor cameras. It needs an alert and a conscious citizen or human interference to cover that piece of the waste thrown on the road by an errant citizen. The caveat here is that the alert and the conscious citizen must have basic knowledge of photography and must be comfortable using apps on smartphones.
Data from both the sensors and the pictures sent by an alert and a conscious citizen are through a complex system of multi-point and multi-layered analytics. We use past waste data to train the system on a machine learning (ML) platform to identify and categorize waste and also approximately to gauge the weight of the waste. The ML platform uses past images of waste taken from over 60 bin locations in the city, at various hours of the day. The ML platform is also trained on regular items generally found in and around waste bins so it can identify them easily.
An IoT-based ML platform for waste management adds a dimension of agile, real-time mapping and tracking through smart use of technology, networking, device or sensor management, and machine automation. Integration of all these needs very little human interference, with most of the activities being automated and monitored round the clock by a smart machine with the ability to analyze pictorial data and to do a bit of number crunching, when needed.