From wireless UUVs and 3D-printed AI aquadrones to bicycles for the visually impaired, this year’s Broadcom MASTERS projects are truly inspiring.
Broadcom MASTERS—the premier STEM competition for middle school students in the United States—is being held virtually from October 22–28. The students’ projects are being showcased on Broadcom MASTERS’ ProjectBoard page from October 26, where visitors are being granted 24/7 access along with the ability to provide feedback within a collaborative environment. (Full disclosure: ProjectBoard is developed and owned by engineering.com.)
Engineering.com had the opportunity to chat with four of Broadcom MASTERS’ 30 finalists:
- Lucas Katz, who designed an underwater system that uses a high-powered LED light to transfer data and transmit energy for remotely charging batteries.
- Camellia Sharma, who built a 3D-printed aerial drone/boat that can take underwater photos and count different species of fish.
- Ansh Sehgal and Avi Patel, who partnered up to develop a wireless bicycle-to-bicycle guide system that allows a visually impaired cyclist to ride with a sense of freedom.
Lucas Katz: Underwater Wireless Transmission of Data and Energy Using Artificial Light
Katz’ project addresses versatile and remote underwater energy and data transfer for unmanned underwater vehicles (UUVs) and floats, which are commonly used for ocean exploration.
“Autonomously connecting a UUV to an electrical cable is difficult underwater, if not impossible,” said Katz. “A method is needed to charge and communicate wirelessly from a distance without having to be electrically connected.”
Katz used artificial light and a floating solar base station to wirelessly transmit data and energy underwater. He first performed experiments to better understand light transmission through water, by coupling different lengths of PVC tubing between an LED light source at the top of the rig and a solar cell at the bottom (and in other cases, a custom-built Arduino luxmeter). Two LEDs were hooked in parallel and connected to a power supply through a voltage divider that Katz engineered. Next, Katz developed a prototype test system. He used Fusion 360 to replicate the main body and attach all components and electronics, which he subsequently built based on his design.
With the pandemic shortening his school hours and allowing him more time to focus on his project, Katz’ inspirations came primarily from YouTube videos about science and engineering. “Some of my favorite YouTube engineers are The Hacksmith, Allen Pan, William Osman, Mark Rober, Peter Sripol, Jake Laser, Tom Stanton, Michael Reeves, The King of Random, and many more,” revealed Katz.
Katz hopes his project would be useful for Seabed 2030, a project to map the entire ocean floor by 2030. “Collecting data about the ocean is critical for understanding its physical properties such as temperature, salinity and acidification, while also creating a better understanding of circulation, plastic pollution, sea level rise, tsunamis, climate change and ecosystems,” said Katz.
We asked Katz if he had any advice for future generations of STEM students. “I suggest that you do what you love, and have your projects reflect on [your interests],” expressed Katz. “If you love sports, do a sports-related project. If you love skateboarding, work on a skateboarding project.”
Camellia Sharma: 3D-Printed AquaDrone and FishPopAI
In another ocean-related (but vastly different) project, Sharma was on a mission to draw attention to stock assessment—a surprisingly overlooked branch of marine science that deals with monitoring the number and type of fish.
“When I started exploring fish stock assessment, I found that stock is currently estimated anecdotally from sources such as publications and fish caught annually,” said Sharma. “Compared to the rapid advances in AI-enabled facial recognition, stock assessment is in infancy.”
As Sharma started designing her FishPopAI software, she discovered that counting and categorizing would require fundamentally different approaches. She chose computer vision (CV) for counting, and artificial intelligence (AI) to categorize different types of fish. (While blob analysis is often used where grayscale images of fish are converted to blobs, this is only suitable for aquaculture and fishery farms where each tank contains a single species of fish.)
Sharma had what she called the “hairbrained” idea to convert the fish profiles into simple geometric primitives such as ellipses and rectangles for processing through AI, using a signature set of primitives to uniquely identify each fish model. (“Much to my amazement and delight, it worked!”) Sharma then tested numerous AI classifiers including convolutional neural network, K-nearest neighbors, random forest, and support vector machine. Her final solution had an accuracy of over 94 percent.
Sharma went on to invent AquaDrone, which combines the capabilities of a remote-controlled boat and aerial drone—and has an additional downward-looking camera in the hull for underwater photography. She is currently in the process of patenting her technology, which she believes has the potential to be used for search and rescue, diverse marine research of aquatic fauna and flora, surveillance, military applications and more.
“Take the example of exploring the ecosystems in the lakes of Minnesota,” explained Sharma. “Investigation of individual lakes yields tremendous value. However, studying them as a set will be far more valuable. Most of these lakes, especially their finger-like stretches, are inaccessible by road. AquaDrone could fly from one lake to another, land on water and record underwater videos. Combined with GPS data, now we have a scope for studying how each microenvironment is working within the context of a wider region.”
Sharma’s family was a big part of her project. With the pandemic limiting access to labs, her dad and brother became her mentors—especially when it came to the steep learning curve for CV and AI algorithms. Inspiration for Sharma’s project came from boat trips with her grandfather, with whom she often fishes and visits remote islands in the Chesapeake Bay. “He told me about the critically declining marine life in the bay and difficulty of studying habitats, especially in the remote estuaries,” conveyed Sharma.
As for Sharma’s words of wisdom, she advised STEM students to take a leap of faith. “People may try to belittle your idea because they think it is too nonsensical, but the best innovations come from those crazy ideas. Also, don’t fear failure. Every single person who has accomplished something great has faced many missteps in their journey. The important thing is to learn from those mistakes going forward.”
Ansh Sehgal and Avi Patel: Bicycle-to-Bicycle System for the Visually Impaired
Back on land, Sehgal and Patel designed and prototyped a semi-autonomous system involving a guided bicycle that wirelessly receives instructions from a pilot bike. The aim was to allow the visually impaired to ride safely without being limited to tandem bikes, through the usage of readily available off-the-shelf sensors.
The pilot bike—ridden by a sighted person—features a data transmitter consisting of a Gyro sensor and GPS, which at each turn sends the yaw angle of the bike and GPS coordinates to the guided bike via a radio frequency (RF) chip.
Three modules are mounted on the guided bike: a data receiver, mechanical clicker and safety module. The data receiver consists of an RF chip receiving data from the pilot bike, a GPS to provide the guided bike’s location, and an MP3 player providing voice instructions. Arduino code calculates the distance, time, and turn angle between the two bikes, which are translated to voice instructions. The mechanical clicker on the bike’s handlebar—made of LEGO Technic parts—produces clicking sounds and haptic sensations based on the degree of the turn, with each click representing 15 degrees.
The safety module comprises three ultrasonic sensors on the front of the bike and two vibrating motors attached to the handle grips. The ultrasonic sensors sense any object within a 10ft distance and 45-degree conical radius, and vibrate the bike’s handle grips to alert the rider of any sudden obstacles.
“Initially, Avi and I designed the electronic modules using Arduino, breadboards, and sensors,” said Sehgal. “We later improved our design by using tools like Fritzing and Autodesk Fusion 360.”
“In the future, we are planning to use Raspberry Pi with a Broadcom chip instead of an Arduino board, for simplifying work and removing complex voltage issues,” added Patel.
Patel’s inspiration for the project stemmed from his time volunteering at a summer camp for visually impaired children when he was 10 years old. During his experience, a tricycle caught his eye and he realized that he hadn’t seen any regular bikes around. This planted a seed which would later blossom in 6th grade when Patel would learn of a local science fair and partner with Sehgal on the project.
During the engineering.com interview, the duo stated that COVID-19 had actually positively impacted their project because “sports got canceled”, allowing them to boost productivity on their STEM project. Sehgal and Patel further asserted that they had access to outstanding mentorship through a program at their local science fair.
As for words of encouragement, the pair said the following: “The school curriculum can only take you so far into the STEM field. STEM is all about diving deep into the scientific process, and having an ever-changing curiosity. No matter what we do, we should always have that curious mind on how things work, and self-motivation to go a step further to learn something new. Tim Minchin said, ‘Science is simply the word we use to describe a method of organizing our curiosity’—and he was a comedian, not a scientist. Our goal as the future generation of STEM is to continue being curious, and never stop learning.”
Engineering.com would like to congratulate this year’s Broadcom MASTERS finalists for their inspiring work in the world of science. To explore all 30 finalists’ projects, click here.
