What are your pet peeves as a person living in IT City Bengaluru? Traffic and trash. We will see how tech can tackle the latter, in this article. According to an article in Deccan Herald, Bengaluru currently generates around 4,300 metric tonnes of waste both from commercial establishments and residential units. This was also reflected in the recent Swachh Survekshan ranking released by the Ministry of Urban Affairs. Bengaluru’s ranking slipped from 28 in 2021 to 43 in 2022. 

SwachBin – the smart bin developed as part of Call for Code Global Challenge – won the fourth runner-up prize in 2022. The winning team members – Suneetha Jonadula and Mohamed Fazil were recognised by Call for Code and were also invited to UNI’s Delegates Dining Room to showcase the innovation.

“Solid Waste Management (SWM) is among the basic essential services provided by municipal authorities in the country to keep urban centers clean. However, municipal bodies are increasingly getting overwhelmed and end up depositing solid waste at dump yards within or outside the city haphazardly,” said Bharathi Athinarayanan, in an exclusive interaction with Analytics India Magazine.  

The team behind the smart bin

Athinarayanan and his team are the brains behind SwachBin, a smart AI-powered trash bin that solves the waste/trash segregation menace at source in a seamless manner. The team comprises of Bharathi Athinarayanan, from Bengaluru, who is the product owner & AI/ML architect; Suneetha Jonadula, from Hyderabad, the lead fullstack developer; Prashanth Parthasarathy, from Bengaluru, the principal application developer; and Mohammed Fazil, from Bengaluru, their AI/ML development engineer.

This smart trash bin identifies the trash and performs classification based on an AI algorithm. The all-in-one bin can collect and segregate five types of trash such as recyclable (wet and dry), and non-recyclable (hazardous, sanitary, construction, and debris). That’s not all, it accomplishes the said task at ultra-low cost and by enabling sustainability. “Through our solution, we are automating the segregation process, which is more efficient than the manual processes,” said Prashanth Parthasarathy, who is part of the team that developed the smartbin.

The team also partnered with BBMP –  Bengaluru’s municipal governing body – that helped the team in identifying the types of trash that the citizens need to segregate and using their inputs the team created their own ‘real’ dataset to be used for the smartbin. What makes it even more efficient and accessible is its low cost of production. “It cost us just around Rs 3,000 to develop the prototype as the hardware is easily available. This model can easily be replicated by others. When produced at scale, the cost will come down to Rs 300 per bin, as the cost of the hardware components is very affordable,” said Mohammed Fazil. 

How does SwachBin work?

SwachBin is equipped with Raspberry Pi and is connected to a camera that takes a picture of the trash. With the help of computer vision, the system predicts the trash and classifies the image accordingly. Based on the result, an electrical signal is sent to the servo motor to rotate which in turn opens the trash can lid either to the left or right, depending on the recyclable or non-recyclable classification.

Once the waste is segregated within the bin, the level of trash is monitored in real time by an ultrasonic sensor. Upon reaching the ‘set’ threshold, the ultrasonic sensor notifies the municipal corporation so they can plan a pick up for the trash. “The IOT dashboard shows how full the bin is. Once filled up to 80% capacity, it sends a notification to BBMP for collection,” said Athinarayanan.

“Due to the infusion of AI/ML and advanced hardware, it is possible to detect the type of trash with high accuracy and also through the use of NLP (text-to-speech), we can educate the user on the type of waste they are disposing of. Upon research, we found no existing solution that would be a combination of high-tech and yet highly affordable in this segment,” he said.

The steps involved

1. The waste material/trash is presented in front of the camera that is connected to the SwachBin (powered by Raspberry Pi).

2. The captured image of the waste is sent to the Raspberry Pi.

3. The AI engine processes the image and identifies the class of the trash and the category it belongs to. SwachBin uses the Resnet algorithm that has a very deep network of up to 152 layers by learning the residual representation functions instead of learning the signal representation directly.

4. Depending on the classification, the AI engine sends the corresponding signal to the servo motor to open the respective lid of the bin.

5. Classification details are further stored in the SQL database. 

6. Activity details get updated in the Docker container.

7. Depending on the class type of the trash, the LED indicator is turned ON and the speaker conveys the same information as a voice message to the user.

8. The ultrasonic sensor senses the trash level and sends the information to the flask app to be displayed in the dashboard.

9. All the information gets displayed in the IoT dashboard.

“In the next six months, we plan to further develop the model by integrating real time regional language translations to help the common citizen,” said Athinarayanan.