Industry-Sponsored Student Capstone Projects

Expendable conductivity-temperature-depth profilers (XCTDs) record salinity and temperature profiles across depths to assess ocean conditions. These single-use devices often cost thousands of dollars per unit, creating an economic barrier for oceanographic research. This team developed Ion-XCTD: a lower-cost platform that preserves core XCTD functionality, integrating UW-APL’s recently developed ion-potential-based salinity sensor, a depth sensor, and a temperature probe. Ion-XCTD houses its embedded system in a waterproof enclosure and collects data as it descends via a spooled tether. The collected data is sent up the tether to a floating buoy where, it is wirelessly relayed to a remote receiver.

The SHEW project seeks to develop a prototype system for harvesting electrical power by utilizing the collection of rainwater in conjunction with a hydro-turbine. With the help of UW CEE, among others, this project aims to design and test the feasibility of a compact hydroelectric system. The prototype consists of a rainwater collection tube and several systems that automate the release of water to drive a turbine. With the capability of producing energy for small and large-scale urban buildings, the system will store power for future use by smaller devices. The power produced can be used to support critical systems such as data centers and communication networks during power outages.

In this project hydrogen generation through an aluminum-water splitting process was pursued as a clean energy option for transportation applications such as watercraft and submersibles, with a need to improve reaction control, propulsion integration, and economic viability beyond an earlier prototype. The student team worked to develop a mini-submersible or watercraft platform that uses hydrogen produced onboard to power a throttleable propulsion system. The project focused on studying the factors that influenced hydrogen production rate, creating a reaction chamber for controlled experiments, and designing a propulsion approach with sound fluid dynamic and material considerations to manage both gas flow and excess heat from hydrogen generation. The team produced a prototype that could demonstrate adjustable hydrogen production and variable-speed propulsion, enabling evaluation of this energy approach for practical marine use, including potential operation in seawater and assessment of its commercial feasibility.

Every month, 10-15 pieces of medical equipment go missing at UW Medical Center because current tracking devices lose power and go silent, which costs the hospital $30 per replacement and hours of staff time. This team built a smarter, smaller asset tracker with a long-lasting battery, motion-based smart wakeup, and a live dashboard that sends maintenance alerts so staff always know where every device is. This team turned a chronic, expensive problem into a solution that hospitals can actually use to track these expensive tools.

This project developed a reliable, external testbench for the UW Medical Cyclotron Facility (UWMCF). The testbench is meant to serve as a safe environment to test updates for the Programmable Logic Controller (PLC) system before pushing them to the main system. The team focused on compiling a Standard Operating Procedure (SOP) for the software and hardware components of the test bench and trained the UWMCF staff on using the testbench environment. By building a separate, independently functioning PLC testbench, where new hardware and software designs can be implemented and tested, plant reliability will be increased and cyclotron downtime will stay minimal.

GASLEAD (Gas Leak Early Anomaly Detection) is an autonomous vibration-based sensing system designed to detect compressed gas leaks in industrial manufacturing environments. Sensor nodes clamp onto compressed air lines, continuously capturing high-frequency pipe vibration data and transmitting it wirelessly over a LoRa network to a central hub. A machine learning model analyzes the data and identifies anomalous vibration signatures indicative of turbulent flow caused by leaks. Unlike traditional methods, GASLEAD operates non-intrusively without requiring system shutdowns or manual inspection, helping facilities reduce energy waste, lower operational costs, and improve workplace safety.

Wide-angle lenses bend and stretch images, which makes them a poor choice for astronomers studying and taking accurate data measurements for meteor observations. This project built a small, battery-powered cart that slides a camera back and forth along a track driven by magnets, letting the camera use a regular lens to capture a wider view. The cart is moved by a reliable custom magnet and coil system using very little battery power. A small computer keeps the camera’s position accurate to within a millimeter and lets the user control everything wirelessly from a phone or laptop.

Meteor entries deliver both atmospheric physics and impact hazards, yet existing networks reconstruct trajectories with straight-line fits that discard curvature and fragmentation signatures. This student team developed a three-station optical triangulation platform that reconstructs curved-path trajectories. Each station streams 4K monochrome imagery at 30 fps through a 130° fisheye lens with sub-millisecond GPS timestamps. A software ring buffer preserves 10 seconds of pre-trigger frames before each detected event. At 30 km baselines, the network triangulates 70% of usable sky and resolves 0.3° curvature on bright fireballs, enabling deceleration profiling and fragmentation-induced lateral velocities.

The IonoSense System provides thermal and humidity management for sensitive signal measuring equipment in harsh outdoor environments. It uses two physically isolated enclosures for its analog and digital subsystems. The analog enclosure protects against EMI and maintains low-noise operation, while the digital enclosure handles data acquisition and active thermal control using infrared temperature sensor readings to drive PWM fans and polyimide heaters for stable internal conditions. The SHT31 humidity sensor serves as an excess humidity alarm. Signal integrity is protected through copper shielding and filtered connectors. Applications include space weather monitoring and geophysical sensing.

The existing crossing of Thornton Creek at State Route (SR) 522 was identified as a barrier to fish passage. To address this, the Washington State Department of Transportation (WSDOT) advanced a project to restore fish passage at this location (milepost 2.86) within WSDOT’s Northwest Region. Building on a preliminary hydraulic design, the project refined a final design concept consistent with the WSDOT Hydraulics Manual and fish-passage requirements, with attention to long-term channel stability and performance over the life of the structure. The work supported WSDOT’s obligations under the federal fish-passage injunction, its commitments related to Tribal Treaty Rights, and coordination with tribal fishery managers and WDFW. Correcting the barrier is expected to provide access to approximately 23,215 linear feet of habitat gain while also exploring opportunities to reduce costs, accelerate construction, and improve channel performance over time.

WSDOT sought a data-informed way to refine where and how speed cameras are deployed in active work zones as part of its speed camera pilot program, launched in partnership with Washington State Patrol and other agencies. This project supported that effort by developing a framework for evaluating camera siting and deployment practices using field observations, site characteristics, speed data, crash history, and before-and-after comparisons of active and inactive camera operations. The work also considered whether driver-behavior indicators, such as acceleration, hard braking, trajectory patterns, and near-miss or telematics data, could help assess work zone safety beyond traditional speed measures. The resulting framework was intended to help WSDOT identify deployment practices that improve its ability to evaluate and predict the safety performance of work zone speed cameras as the pilot program continues to unfold.

This project focused on redesigning an aquatic wheelchair based on the current wheelchairs in use at the Bellevue Aquatic Center. The redesign needed to be cost-effective, reproducible, durable, functional, and comfortable to be used in a warm water therapy pool on a zero entry ramp. It also needed to fit on the pool ramp, through 36-inch doorways, and between the pool, locker rooms, and showers. The wheelchair needed to support up to 400 pounds, roll and turn smoothly with a seated user, provide stable support and relative comfort for nonambulatory users, reduce excessive buoyancy in the water, keep the user positioned safely as the ramp descends to 3 feet, and include functional wheel locks on both main wheels. The project resulted in open-source aquatic wheelchair plans and a prototype intended for use at the Bellevue Aquatic Center.

Wyze needed a clearer way to understand how customers moved through a complex app experience that generated more than 2,000 distinct event types across device setup, settings, content viewing, subscriptions, and feature use. The project developed a structured event taxonomy and customer journey mapping capability using app analytics and warehouse data to organize events into consistent categories, define shared meanings, and tag interactions by journey stage such as onboarding, activation, engagement, support, monetization, and retention. It also outlined critical user paths by analyzing event sequences, funnels, and drop-off points, with segmentation by factors such as device type, subscription status, and feature usage. This work was intended to give Wyze a governed framework for interpreting app behavior, align teams on common journey definitions and key metrics, and support more consistent product measurement, experimentation, and decisions about onboarding, engagement, and monetization.

Wyze needed a way to improve the accuracy and consistency of Zendesk ticket tagging, which had been time consuming to do manually and often varied across agents and support partners. The project developed an AI-assisted classification capability that used ticket comments together with call or chat transcripts to propose the appropriate tags and dropdown field values in real time, with an option to automatically apply selections when confidence was high. Integrated with Zendesk, the system was intended to surface suggestions, confidence levels, and brief explanations within the ticket workflow while capturing overrides for review. It also included evaluation, monitoring, audit logging, and privacy safeguards such as PII redaction and controlled logging. This capability was designed to reduce agent effort, improve data quality for routing and analytics, and support more consistent taxonomy use across support channels.