Progress:
- Wes: Designed the power subsystem (battery sizing, power connectors) and quantified the mic array’s design requirements (SNR, aperture, sound pressure math) in the Design Report. Finalized PCB schematic/layout in KiCad, passed DRC, got a fabrication quote (~$30/5 boards).
- Roshni: Implemented the bandpass filter in SystemVerilog (previously Python-only), built automated tests comparing hardware vs. software output, found and fixed two bugs (scaling error, pipeline timing), now matches the Python model exactly.
- Essie: Stress-tested the heatmap against noise/multiple leaks, found and fixed a false-positive bug, reduced the error rate from 12% to 1.5% (under the 10% requirement) via averaging.
Risks: Several teams’ numbers (array geometry, filter band, noise thresholds) are still placeholders pending real hardware — mitigated by developing independently against realistic assumptions now, syncing once hardware’s in hand.
Design changes: None.
Schedule: On track, all three members.
Highlights: Wes has a fabrication-ready PCB with a real quote; Roshni’s RTL now matches the software model sample-for-sample; Essie’s fix brought the false-positive rate from 12% to 1.5%.
Part A Public Health and Safety
SoundSight reduces physiological risk by letting technicians detect gas leaks, including hazardous or asphyxiating gases without needing prolonged close-range exposure to the leak source. The device is non-contact and passive, it localizes leaks acoustically from up to 10 feet away, so the technician doesn’t need to physically approach, touch, or probe potentially pressurized or hazardous equipment the way contact-based methods require. This directly removes a physical-harm pathway from the inspection process itself.
Part A was written by Essie.
Part B Social Factors:
Project SoundSight addresses the social barrier created by the high cost of existing ultrasonic acoustic cameras, which can run tens of thousands of dollars and are therefore accessible only to large corporations with dedicated reliability teams. By building on a low-cost FPGA platform and commodity MEMS microphones, the device lowers the economic threshold so that smaller manufacturers, municipal utilities, public school districts maintaining aging HVAC systems, and community-owned cooperatives can perform inspections that were previously out of reach. This broadens who gets to participate in proactive infrastructure maintenance rather than confining it to well-funded specialists.
Part B was written by Wes.
Part C Economic Factors:
Our product targets an economic problem for facilities that run compressed air or pressurized gas lines: leaks waste energy and money, and finding them takes time. The U.S. Department of Energy estimates that leaks can waste up to 20 to 30 percent of a compressed-air system’s output, which is a continuing operating cost for any plant that runs one. Routine leak surveys are how facilities control that cost, but the tools available sit at two extremes. Professional acoustic-imaging cameras (FLIR’s Si-series, Distran) are priced for large industrial budgets. Fixed camera systems like the one in Lee et al. (2024) need dedicated hardware at each monitored point, so covering a whole facility would take one unit per fitting. Smaller facilities and budget-limited maintenance teams often end up skipping routine surveys, and they absorb the wasted energy as a result.
Part C was written by Roshni.