The Robotic System Combating the Soft Plastic Recycling Crisis

Why is the sorting of waste essential? And what are some of the limitations with current systems?

The sorting of waste is central at material recovery facilities (MRFs). These are the factories that process our recyclable waste. Waste needs to be sorted as industries that use recycled waste as a material for manufacturing have waste contamination requirements. If they require plastics for their manufacturing processes, the plastic cannot be mixed with, paper or metal or glass. If the material is contaminated, it cannot be used and it ends up in landfills.

They first need to separate waste to gather material, which can be used for further processing, with no contamination. If the material is contaminated, then it is thrown away.

Image Credit: The University of Sydney

Can you explain what some of the problems are with the recycling of soft plastics?

It's very hard to recycle soft plastic because it is hard to distinguish it from paper.

Also, when processing soft plastics, they can get into machinery and cause blockages. Soft plastics can get entangled into machines and can cause failures, resulting in factory downtime. As a result, only a very small percentage of soft plastics is processed, around 6%, and current technology solutions are inadequate.

How could robotics solve these issues?

The system that we developed is not only based on robotics, it is the first system in the world with an IoT system. It is important to note that the current solutions for recycling soft plastics are mainly based on human sorting. We were inspired to make a process similar to how a human works. So we made an IoT system that had similar operations to a human. We have 3D cameras that serve as the human eye, communication networks that serve as human nerves, AI as the human brain, and a robot arm as the human arm.

We integrate all these systems together to make a solution that can operate like a human. We believe this is the first integration of these technologies into such a system to provide enough accuracy and speed in processing soft plastics.

How does the robotic automation system see and sort the waste?

The system sees through 3D cameras, which are combined into computer vision with AI. It is a self-learning process, so when they see enough pictures, they can recognize what they need to identify. It is a system that is supported by machine learning.

AI is also involved in training the robot to pick up objects precisely, using appropriate angles and speed; the speed has to be synchronized with the movement of the conveyor belt.

Can you describe the stages of the sorting process?

This process is combined with household collection. The process is organized to be a collaboration between councils and material recovery facilities. Councils will distribute some special bags to households, then people will collect soft plastics into these special bags.

These bags will come through recycling bins and will be distributed to MRFs together with other commingled waste. The commingled waste will come through on the conveyor belt and robots will recognize them by the process that I've described. The robots will then separate them into multiple streams, for further processing. So, that could be shredding and grading and preparing to make fuel and chemical products, which would be suitable for making plastics again for packaging or making indoor-outdoor furniture or signage.

The robot will separate different types of plastic. Credit: University of Sydney

Can you explain who some of the contributors to this research were?

The contributors are Dr. Wanchun Liu, a postdoc in our lab, with expertise in networked control, Mr. Dawei Tan, a senior technical officer, an expert in hardware and software design. Professor Wanli Ouyang is an expert in computer vision, and Professor Yonghui Li is an expert in telecommunications and IoT. My role was to lead the whole project and coordinate all these operations.

What was the next stage for the project?

The next stage is the deployment in factories. It took us around one year to develop the project, and then we applied for a government grant. Last year a grant was approved and we negotiated the contract for a few months.

Soon we will deploy the laboratory solution in factories. And we also plan to expand the technology for processing other types of waste, for example, sorting other types of plastics and used clothes.

Should pilot schemes such as using the system in IQ Renew's material recovery facility be successful, do you envision this to be a widely adopted system by the waste management and recycling industries?

Yes, that's part of this project, a three-year project, for which we just got the funding of $3 million for the three years. As part of the project, there will be a package designed that will be offered and distributed to other MRFs to convert them into so-called smart MRFs or S-MRFs. So, that means deploying the technology at the national level.

How does this system compare to current measures in terms of operating costs?

Previously, soft plastics could only be sorted by humans, so in terms of cost and productivity, the system we have developed will be much more cost-effective and productive. For example, robots can have 2000 picks per hour, while humans can have a maximum of 800 picks. It is also a system that is much easier to scale.

Is this a scalable system?

Yes, this is a scalable system because its main components are implemented in software, which could be easily deployed.

Are there any improvements to the system that the partners are currently working on?

We are working on improving the speed and the accuracy of the system. So, working on getting past robots and improving the accuracy of waste detection, base type detection and improving the robot reaping operation.

What's next for the team at the University of Sydney?

We are working on communications, which is the central part of IoT; that's our area of expertise. We've been working on developing 4G, and 5G, which is being deployed worldwide now. But now, the next research aim is to work on 6G, which will be deployed after 2030, so that's one of the current main research tasks.

We are also developing IoT for other applications. In health care, for example, we have developed a system with ECGs, the system sends signals that the ECG generated, via communication networks, or the cloud with AI capabilities, where the AI engine can automatically diagnose various diseases.

We've also been working on the application of IoT in the automation of warehouses for e-commerce, for example, and automation in agriculture for meat processing and meat tracing and the detection of various microbes. We have also been working on the application of IoT for the automation of warehouses in supermarkets.

Where can readers find more information?

More information can be found on our website.

About Branka Vucetic

Professor Branka Vucetic's work aims to develop theoretical framework and design principles for wireless communication systems. She is an internationally recognized expert in coding theory and its applications in wireless engineering. Professor Vucetic has held various research and academic positions in the UK, Yugoslavia and Australia. Since 1986 she has been with the School of Electrical and Information Engineering at Sydney University, where she is currently Laureate Professor and Director of the Centre of Excellence in Telecommunications.

Her research interests include wireless communications, digital communication theory, error control coding and multi-user detection. Prof Vucetic published four books and more than three hundred papers in telecommunications journals and conference proceedings.

She is a Fellow to the Australian Academy of Science (AAS), Fellow of the Australian Academy of Technological Sciences and Engineering (ATSE), a Chinese Government Friendship Award recipient, an IEEE Fellow and a former Editor for the IEEE Transactions on Communications. In the last several years she has managed several projects related to wireless communications networks development, addressing the issues like interference cancellation, multiple antenna signal processing and coding, as well as multiple access technologies.

Interview questions provided by Robert Lea.