Teen Develops Flood-Detecting CubeSat – IEEE Spectrum

At present satellites, artificial aperture radar, and GPS are used to gather knowledge on flood harm, monitor the placement of victims, and talk with emergency responders. However know-how failures and sluggish knowledge transmission speeds result in delays in response time, Service provider says. The rise in international flooding has intensified the necessity for extra correct and dependable strategies.

Final yr Service provider constructed what she says is a more practical approach to monitor and acquire knowledge throughout floods: a small, cheap, standardized CubeSat built-in with synthetic intelligence. The little satellites use a a number of of 10- by 10- by 10-centimeter models—which permits producers to develop their batteries, photo voltaic panels, computer systems, and different elements as off-the-shelf elements.

The CubeSat takes pictures of an space and makes use of sample recognition to detect flooding, assess infrastructure harm, and monitor survivors.

Service provider offered her paper on the machine at this yr’s IEEE Area 3 annual convention, IEEE SoutheastCon.

“IEEE is a foundational a part of my development as a younger researcher,” she says. “It turned engineering from my dream to actuality.”

Constructing a CubeSat at MIT

Service provider says her curiosity in catastrophe response was sparked after studying that it will possibly take a number of hours for emergency employees to obtain satellite tv for pc knowledge.

Decided to discover a quicker technique, she started researching applied sciences and found what CubeSats can do.

“CubeSats are very agile, scalable, and able to forming constellations (multiple-satellite teams) that replace knowledge in almost actual time,” she says. “The concept these small satellites—which match into the palm of your hand—might ship life-saving insights quicker than conventional techniques actually impressed me to push the idea additional.”

Final yr Service provider and three of her classmates have been accepted into MIT’s Beaver Works Construct a CubeSat Problem, the place groups of as much as 5 U.S. highschool college students got eight months to develop a satellite tv for pc able to finishing a space-based analysis mission.

Service provider’s group—the Satellite tv for pc Sentinels—constructed a CubeSat powered by a convolutional neural community (CNN) that may establish closely impacted flood zones and remotely acquire knowledge for catastrophe reduction and environmental monitoring. CNNs analyze picture knowledge for sample recognition.

Service provider was the group’s payload programmer and led the mission’s design and simulation efforts, which included planning, configuring {hardware}, and creating autonomous software program and algorithms to handle the payload.

The group started by making a 3D mannequin of the machine to visualise and refine the position of its elements. The know-how used—together with a Raspberry Pi, a number of sensors, and a digital camera—was housed in a transparent plastic dice.

The center CubeSat was developed by Service provider and her group throughout the MIT Beaver Works Construct a CubeSat Problem. On the left is a business 1U CubeSat whereas on the precise is a prototype of Service provider’s present design. Abigail Service provider

The machine, which value US $310 to construct, weighs about 495 grams and was remotely related to a laptop computer through Bluetooth throughout ground-based testing. The pc incorporates a machine studying algorithm—written by Service provider utilizing Python—that analyzes collected pictures to detect flooding.

The CubeSat takes a high-definition picture of its environment each 2 minutes and transmits it to the laptop computer. The satellite tv for pc transfers as much as 1,500 pictures day by day and shops them on a 16-gigabyte SD card.

The algorithm then analyzes patterns, together with adjustments within the water’s shade and the picture’s pixel density. When the algorithm detects flooding, the machine can alert emergency responders.

“Whereas many present techniques function on multihour cycles, the CubeSat captures high-resolution pictures each 2 minutes,” Service provider says. “The system can then set off alerts which can be delivered to first responders through SMS or electronic mail.”

To check their system, Service provider and her group constructed a metropolis mannequin manufactured from Lego blocks in an empty bathtub. They positioned the CubeSat over it, and it took pictures of the scene. They then added water and dust to make it look extra like an actual flood. The CubeSat efficiently transferred the pictures to the laptop computer, and the algorithm detected the flooding.

Out of 30 groups, the Satellite tv for pc Sentinels positioned third.

Persevering with her work at Accenture

Service provider is constant her analysis on flood-prevention applied sciences at Accenture in Richmond, Va., the place she works remotely as a payload proprietor and designer for the corporate’s CubeSat launch group.

After the MIT program ended, Service provider determined to scale her challenge. She reached out to her former mentor Chris Hudson, the worldwide technical lead in house cybersecurity at Accenture. He provided her an internship.

Service provider is working to make the transition from prototype to purposeful product however, she says, wants to beat obstacles she encountered along with her MIT challenge.

The primary one was that the mannequin struggled to detect flooding in variable circumstances. It’s as a result of the CNN mannequin wants context, she says. With out it, the mannequin can misread advanced visible cues. To repair the difficulty, Service provider educated the algorithm to identify flooding by figuring out colours in particular person pixels.

Transmitting pictures utilizing Bluetooth labored in her lavatory, nevertheless it isn’t fairly as helpful when CubeSats are orbiting 700 kilometers above the bottom.

“In case you’ve used a Bluetooth headset earlier than, you already know it disconnects the second you stroll away from the machine it’s related to,” she says. “That isn’t going to work when the CubeSat constellation is in orbit.”

She advised the Accenture group swap to SubMiniature Model A (SMA) antennas. The RF antennas hook up with the CubeSats utilizing an SMA connector.

“The event course of has been probably the most formative experiences of my profession to this point,” Service provider says. “Working by way of the payload design and validation and assembly with these groups has given me a lot expertise, particularly for my age.”

Her payload is predicted to be launched early subsequent yr.

An aerospace internship at MIT

Service provider is an intern on the MIT Pc Science and Synthetic Intelligence Laboratory, the varsity’s largest interdisciplinary lab, with 60 analysis teams. CSAIL is led by IEEE Fellow Daniela Rus, recipient of the 2025 IEEE Edison Medal.

The internship is distant, and Service provider conducts analysis in a laboratory on the College of Central Florida, in Orlando.

“IEEE is a foundational a part of my development as a younger researcher. It turned engineering from a dream to actuality.”

In that function, Service provider is specializing in cognitive cartography, a way for structuring advanced data into semantic maps that reveal how concepts and ideas relate to at least one one other. She makes use of embedding fashions, a kind of machine studying that converts data into numerical representations. The embeddings enable computer systems to acknowledge similarities and relationships between ideas, even when they’re described in several methods. The method helps an AI product perceive how concepts join, relatively than treating each bit of information as remoted.

“Being one of many youngest individuals within the lab is daunting,” Service provider says. “Nonetheless, I’m actually excited to be taught from engineers and researchers who’re working on the slicing fringe of the sphere.”

She says she is hoping to attend MIT or Stanford.

The way forward for IEEE

Service provider was launched to IEEE by Joe Jusai, former finance chair of the IEEE Orlando Part.

Her first private expertise with the group occurred in 2023 whereas she was conducting analysis for a science truthful challenge. She was engaged on a robotic arm that would decide up objects utilizing an electroencephalogram and Bluetooth. The challenge was impressed by her grandmother, who suffers from mobility points and was wheelchair-bound.

“I saved seeing IEEE talked about in each regulation and normal I discovered,” Service provider says. When she realized about an upcoming Orlando Part assembly, she requested her mom to take her.

On the assembly, a number of members offered their analysis. Service provider requested Masood Ejaz and Varadraj Gurupur—the chapter chair and cochair—if she might talk about her science truthful challenge.

“After presenting my work, IEEE rapidly turned a group that has formed my understanding of what engineering can accomplish,” she says.

She felt on prime of the world, she says, when she offered her paper about her CubeSat challenge at IEEE SouthEastCon.

“It’s a kind of experiences that basically adjustments you,” she says.

She is happy to turn into an IEEE scholar member when she begins school, she says. She additionally has her sights set on being elected as its president sometime.

“I met Kathleen Kramer at one in every of my native IEEE occasions earlier than she was elected IEEE president, and we spoke about my work,” she says. “After she was elected, I spotted that I might like to turn into the president of IEEE sometime.

“I hope sooner or later that I can step into the identical footwear as her and proceed to assist IEEE the identical manner it helped me.”