EPICS in IEEE checked in with the student team running the Dreck Administration for Immaculate Purlieu project—and they’re well on the way to achieving their project goals of helping the environment. All members are aspiring engineers, who are thrilled by the hands-on learning experience that they have embarked on this year. 

Working with the IEEE Student Branch of Vellore Institute of Technology Chennai, university students have come up with an easily manageable and cost-efficient system for trash collection. The group includes 5 student leads [Vaishnavi Dixit, Makesh Srinivasan, Dhriti Rajani, Sreedutt Haridos & Sai Samyukhta N] and 5 volunteer collaborators [Arnav Maheshwari, Nishant Kumar, Ronith Jaju, Sanyam Khaiwal & Vidhi Shah].

“The team is creating a smartphone app that will connect to a sensor placed on our improved trash can design. The app will display a red trash can icon when it’s full or a green icon when it’s empty. Based on this coloring mechanism, the garbage truck can check the app and pick up the garbage accordingly,” says team member Sai Samyuktha. The goal is that the municipality will use this information to prevent the trash from being left behind while also saving time and fuel because the workers do not need to physically check all the bins. Additionally, the team built a working robot that will collect litter from the side of the street. 

The team’s next step is to create a smart segregator which will dispose of the waste into one of four different bins depending on the item’s material, such as organic, plastic, metal, and glass. This sorting process will employ machine learning algorithms to efficiently separate the waste after it is dumped into a trash bin. Once this phase is complete, the project’s next steps will pertain to how the trash is disposed of, which could include vegetable waste being turned into fertilizer and polyester being repurposed into new clothing. 

Vaishnavi Dixit says, “This project is a nice way of reaching out to people and enhancing their knowledge and implementing it, within our university and community for a better and sustainable future.” As they work towards accomplishing the many components of their project, the students talked with EPICS in IEEE about their experience with the software and hardware designs, as well as their collaboration and teamwork. 

Hardware

Led by Vaishnavi Dixit, currently, in year two of her Bachelor’s Degree in the Computer Science and Engineering program, the hardware team already has a working robotic garbage car with two robotic arms and one large dustbin attached. Once the robot identifies garbage, it picks it up in an effort to help clean the environment. 

The team’s next step is attaching tracking system sensors that will track and monitor the robot’s location while allowing it to navigate the roads. The robot prototype currently works via remote control but eventually, the sensors will be linked and the team will be able to control the robot with the app that they are creating. The team also plans to design a smart segregator with four bins. These bins will be joined together using a conveyor belt and one waste inlet. sing Machine Learning technology, the object will be detected if it’s dreck or not, and then using the robotic arm it will be picked up and moved to the segregator inlet, where it will be classified it the proper disposal methods.

Although most of the team members had more prior experience with software, they greatly enjoyed working together on the hardware side of the project. “I learned many new things about robotics—and I have a lot more to learn in robotics right now, which fascinates me,” said Dhriti Rajani, a team member and the financial lead for the project. 

Vaishnavi Dixit, who had more experience in robotics, found that she was able to share her knowledge with her teammates while also learning new skills herself. “I have had a lot of experience in hardware but for this project, we had to work on several new software aspects, as a result, the team ended up learning and implying a new set of skills at the same time,” she said. ”It was a totally new experience for me and I got to learn quite a bit.

Software 

For the software implementation aspect of the project, the primary goal is to work on creating a mechanism that can accurately identify and classify various kinds of waste products. The software team’s current objective is to create a model to effectively predict the type of waste based on images of trash. To do this, they have generated a dataset with six classes of images: cardboard, paper, glass, metal, plastic, and miscellaneous. They’re using a Convolutional neural network in order to train to detect waste items and classify them appropriately. 

The model trained over 3000 images is about 85% accurate in determining the waste categories. “We initially got an accuracy of about 60%. However, we felt that we certainly could improve if we had a better and larger set of data. So, we found a way to augment the images that we already have,” said Makesh Srinivasan, the programming lead who served as the previous Vice President and is now an Advisory board member of the IEEE Student Branch of Vellore Institute of Technology, Chennai. He also went on to outline his ideas on the next steps towards creating more prediction classes and exploring various machine learning strategies in improving the existing models.

The team acquired a set of images from various open-source repositories. They sorted the images and saved them into directories named after each of the six item classes.  Using the image generator from Keras API, the team augmented the images and loaded them with a size of 300×300 pixels to ensure the model learns from every possible scenario. The team is also scraping more images from Google Images, as well as generating more classes for identification. 

In the meantime, the students are also developing a more elaborate architecture by using an attention mechanism and skip connections with novel architectures. The software team is now on a mission to increase the size of the dataset and improve the performance of the model to achieve a higher accuracy percentage. The next step is to collaborate with the hardware team to integrate and implement the same in the garbage bins. As ardent and curious are the members are in machine learning, they also plan to work on a smartphone application to remotely observe and control the bins eventually.

Teamwork Leading to Community Impact

These students are learning so much by working together in a multidisciplinary team. They explained that it was tough at the beginning when they had no choice but to meet virtually. When they were able to meet in person, they discovered that they found more enjoyment in collaborating and learning things that they did not know before. 

“We’re all from different years and different disciplinary backgrounds, so it was only a couple of months ago that we all met in person,” said Makesh Srinivasan. “I think this project is one of the greatest opportunities we’ve had in our college career,” he continues. “The opportunity to work on something that we are all passionate about while also benefiting society feels like an amalgamation of everything that we have learned and worked towards.”

It was clear from speaking with the team members that not only were they interested in the project but they were also very passionate about the end goal to help improve sustainability in their community. With the success of their project, the team believes that their solution could result in clean roads, clean bodies of water, and less waste generation in villages—in a more economical way than the types of equipment used for the Dreck Administration today. 

“Now more than ever the concept of sustainability is important. And it makes me really happy to see that investment in sustainability is growing in popularity day by day,” said Sreedut Haridos, publicity lead for the project. “We’re moving towards a common goal of preserving our earth and leaving it better than ever for all. I just want to say we must always remember that we have just one world, and we need to heal it—and that starts with us.”

They went on to discuss that the team has started an Instagram page for promoting their project. They will also look into starting a Linkedin page and using videos of the project prototype to grow interest in the project. The team is also planning to send a survey to gather feedback from the local community in order to better understand the community’s needs. 

Once they begin the next phase and have more information to share, they will be reaching out to their community partner, SHUDDHI™, which is a registered NGO ​working together with partners and ​local communities in India and globally to improve the environment and human well-being. This project is focused on creating a more innovative and effective waste management solution to help protect the Earth from drastic climate change as well as pollution. They plan to start implementing it within nearby villages, with the wide-scale goal of extending it to other states and countries.  

The team has learned a lot during this project so far including, software and hardware design, communication, and budget management. Sreedutt went on to say, “I think it was very different from my engineering experience with respect to this project. It felt like I was learning a lot more, and it has been an amazing experience.” The EPICS in IEEE committee is excited by the progress made so far by the group, and we’re looking forward to the results of the project.