At Wild Planet, we believe that it is important for kids everywhere to learn to care and protect Earth. Anna Du is a teenager doing just that; she’s developed a robot that can detect microplastics in water. In this interview, Anna explains how she created this robot and what she has planned for the future.
1. Can you tell us a little about yourself?
My name is Anna Du, I’m 13 years old, and I live in Andover, Massachusetts. For the past two years, I’ve been working on a remotely operating vehicle (ROV) that uses artificial intelligence to identify microplastics on the ocean floor. I am currently working on a method to spatially map where microplastics are located on the ocean floor by observing associated environmental factors. This is according to the assumption that nature has a way of sorting itself based on different factors. In the future, I hope to use these spatial maps to help create a plan to clean up the microplastics on the ocean floor by prioritizing what is the most dangerous to the environment.
2. What motivated you to design this robot?
I have always been interested in the environment -- living in Andover means that I am surrounded by nature at all times, which contributed to my love for it. As soon as I heard about this massive issue, I started researching more about it to see if I would be able to understand it more and perhaps create a solution for it. I learned about how these plastics spread throughout the entire ocean, through processes known as biomagnification and bioaccumulation -- causing a major ecosystem catastrophe, and potentially upsetting the delicate balance in the environment. But I soon realized that human lives would be affected by this as well. As microplastics spread throughout the global food web, it will cause a multitude of health concerns for us as well. That is why I designed an ROV that can identify microplastics on the ocean floor.
3. Can you explain how microplastics are different than plastic like a water bottle or straw?
Plastic objects, such as water bottles and straws, are one large structure that is relatively easier to clean up. These objects are all in one area, and most of them are visible to the naked eye. Microplastics, on the other hand, may be the same products, but they have broken down, through biological (such as the fish eating the plastics), physical (the plastics being eroded by wave action), and chemical processes (UV cleaving the plastics) into many pieces that are smaller than 5mm. These microplastics will then enter the ecosystem much more easily and are much harder to take out -- not only because they are smaller, but also because whereas before, there was only one piece of plastic to remove, now there are hundreds, if not thousands.
4. How do microplastics affect our environment and the ocean ecosystem?
Microplastics will enter the environment either already as microplastics, such as from cosmetic products, or as larger objects, such as plastic bags, that have broken down due to smaller pieces due to UV deterioration or wave actions. Because of the size of these plastics, they can enter the food chain undetected, with the animals believing they are eating food. This can cause animals to die of malnutrition because while it feels full, its stomach is filled with microplastics and nothing of any nutritional value. Recently, there has been scientific research that proves that microplastics are correlated to many health concerns, such as lung and heart damage, cancer, and even genetic mutations.
5. How did you learn to create a robot, write a computer program, and learn what type of sensors your robot needed?
In order to create my ROV, I had to learn many different new fields of science, engineering, and math, which hadn’t yet been taught in school. To do this, I spent a lot of time in my local library and maker labs, as well as considerable time writing to various scientists and engineers in my area, who could potentially serve as key mentors. While many of these experts did not have much time to spend with me, nearly everyone I wrote to replied and offered me some encouraging words. A number of them have spent a great deal of time with me over the past two years, and for that, I am very grateful.
In the beginning, my ROV was shaped like a cube -- but eventually, through trial and error, I added legs on the bottom to make it sturdier. One of my mentors, from MIT, advised me to make the arms longer as well, to have better maneuverability. I also spent a great deal of time on my own researching how to create the programs required to run my detection system, largely through trial and error. I had some basic knowledge in Python, but in order to program my systems, I needed to use libraries and certain coding environments that I had never worked with before. In particular, the numerous deep learning and analysis tools utilized by the Tensorflow platform.
After experiencing error message after error message, over the course of many months leading up to each science fair, I eventually was able to master the software and now have a highly accurate system that can achieve average accuracies over 90%. In order to incorporate sensor data, I first looked at various environmental factors in the ocean that might potentially cause or be associated with particle aggregation. I chose to incorporate a wide variety of rugged sensors that could easily integrate with a low-power microcontroller and data storage system; thee sensors included temperature, pressure, turbidity, gyroscopes, light intensity, and several more. In addition, in order to incorporate a low-cost spectrography module, I learned how to build my own custom PCB containing an array of infrared lights as well as an infrared camera. I also spent a considerable amount of time learning how to waterproof individual components, which has been very challenging. I am still continuing to tweak my system as I learn more about how to make improvements; it is my hope to make the system smaller, more energy efficient and streamlined, and eventually fully autonomous.
6. What were some of the obstacles you faced when designing the robot?
One of the biggest obstacles I faced was training the neural networks used by my detection system. Creating an artificially intelligent system isn’t just about writing a program; the system needs to learn, just like a human, using a considerable amount of input from the real world. In the case of my system, the samples required to train the model properly is in the tens of thousands of individual data points. But unlike a human, a machine does not have years to learn these things. I needed to learn how to create data sets to train my system to achieve a high enough accuracy, so the system could work in an actual ocean environment, with minimal false positives. I learned that the process of obtaining a large enough volume of high quality “big data” is a fairly difficult task.
Through research, I discovered that many of the world’s leading artificial intelligence softwares use data augmentation. It took me a considerable amount of time to develop a system capable of emulating all of the various aspects of a marine environment, so that I could create realistic augmented data, to supplement my actual samples. This required a steep learning curve in order to make the data both realistic and as randomized as possible. However, now, rather than taking days, if not weeks, to obtain the tens of thousands of data points I need to conduct a single test, I could so with the click of a single button.
While this process was very unknown and foreign to me at the time, having studied it over the course of several months, I was able to develop a program that worked. Now, I apply data augmentation to all my various AI programs, and I have also made my software available to other researchers who need similar types of particle data augmentation for their systems as well. By doing so, I hope that I am able to reduce this obstacle for others as well.
7. Why did you decide to start writing your book Microplastics and Me?
Ever since I’ve learned about the dangers that microplastics pose to the environment, I’ve wanted to spread the knowledge about it as much as I could. I realized that many people weren’t aware that microplastics were even a problem, or they thought that it wasn’t a big deal. But the fact is, microplastics in the ocean is an issue that is steadily growing and will soon affect us all. Not only that, but I also hope that students my age start to take action against problems in the environment – especially in this day and age, where climate change and pollution is really taking a toll on the ecosystem.
When I was approached by an internationally known STEM publisher called Tumblehome Learning, I thought that this would be the ideal partnership to spread the word about this massive issue, and hopefully, inspire other students around the world to also try to tackle a world problem using hands-on methods and their engineering knowledge.
8. What do you wish people knew about how plastics affect our planet?
Plastic pollution will have many future detrimental effects on the environment. Not only will plastics upset the delicate balance of the marine ecosystem, but they will also have many potentially dangerous effects on human health. When plastic pollution enters the environment, before breaking down, they have the potential to trap or strangle animals. However, as these larger pieces break down into microplastics, they pose an even larger danger to the environment.
When synthetic polymers decompose, they often release precursor chemicals that are embedded within the plastics. Humans are also affected by this massive issue -- there are studies that show that if pregnant women are exposed to these pieces of plastics, their children are more likely to develop genetic mutations. Many people think that there are other global issues, such as climate change and energy-related issues that are more dangerous than plastic pollution but in reality, all of these issues are directly linked to each other, through the common link of overconsumption of fossil fuels. There are numerous ways that people can help to solve this issue through community-based cleanup efforts, participation in citizen science projects, and reducing single-use plastic usage to prevent these plastics from ending up in landfills in the first place. Every day, new solutions are being created, and it is important for people to stay informed.
9. If you could change the way plastics are recycled or handled, what would it be?
A number of months ago, I received a notice from my town’s waste management department. The policy had changed regarding the types of plastics that we are allowed to recycle; many of the plastic types that we previously could recycle were no longer permitted. Most of the plastics that enter the recycling plants don’t get recycled at all -- instead, they end up in landfills.
Essentially only the first two types of plastics, PETE and HDPE are recycled, because they are the easiest to recycle. PETE, which is found in water bottles, is easily shredded, and HDPE, commonly found in packaging such as milk jugs, has a low melting point. By making sure more plastics are recycled, we are using a more ‘upstream’ methods to solve this problem, by preventing the plastics from entering the ocean in the first place. In order to spread the word about this, I have created a new not-for-profit organization, Deep Plastics Initiative, that focuses on inspiring others to help clean our planet and provides free educational resources for people to learn more about the numerous ways to do so. I encourage people to organize their own environmentally-minded peer groups, who can work together to take on the task of recycling plastics, especially the types rejected by recycling centers, within their own communities. This way people can create new value out of recycled products and fewer plastics will end up in landfills or the ocean.
10. What advice do you have for other kids who have an interest in the environment and plastic pollution but aren’t sure where to start?
I would encourage my fellow students to visit local libraries, universities, environmental and other organizations to find out about the pollution issues facing your own community. It is really important to read as much as you can to learn more about these issues. Working on a science project of your own, that focuses on environmental issues - in particular, plastic pollution - is a great way to understand these topics in-depth. Not to mention, there are many projects you can do, that affect your community and the world directly. For example, students could work on recycling more and upcycling pieces of plastics. They could also participate in citizen science projects, community recycling/cleanup efforts, or work with a team to come up with new ways to prevent single-use plastics. But in my opinion, the best method is to do a project on it themselves. This way they learn more about the issue at hand and gain lifelong knowledge about the world and foster an appreciation for the science and engineering skills that will last a lifetime.
11. How can our readers support your robot and book?
It is important to me that my book gets into the hands of as many students as possible. I want to make sure that everybody is aware of the danger that microplastics will pose to our society and to our entire environment. The biggest way someone can help me on my mission to get rid of microplastics is to spread the word about the catastrophic effects it will have in our world. Tell as many people as possible about this issue. Donate copies of my book to your local schools or libraries, especially in high needs areas. The point is to make sure that more people are aware of this issue and are equipped with the knowledge they need to come up with potential solutions to this issue.
12. Is there anything you would like to add?
Throughout this process, I have created many programs to identify the microplastics on the ocean floor. With my unnatural color detection system, I can target the colors which would be indicative of synthetic dyes, and using my morphology classifier, I can identify the shapes that would be common in microplastics, such as fibers, nurdles, angular fragments, and textured polystyrene. My programs have also been able to identify the number of particles on a screen, along with the average size range. Using the data I collected, in conjunction with environmental sensors which can identify the factors that would affect where microplastics would be located, I’ll create a spatial map that can predict where the microplastics will congregate on the ocean floor. My plan is to make all of these programs open source, so everyone can use it to identify microplastics. Not only that, but people can also make their own improvements as well, and we all can work to make these programs more accurate and efficient.