Articles and Research Papers

This study presents a budget-friendly air-quality monitoring setup for school classrooms using ESP32 microcontrollers and optical dust sensors. The author assembled eight nodes, calibrated them against a reference monitor, and logged PM2.5 values over six weeks. The network achieved an average absolute error of 4.8 µg/m³ after calibration and helped identify ventilation periods associated with elevated particle levels. The work demonstrates an approachable pathway for high-school students to conduct meaningful environmental sensing projects.

This study evaluates a lightweight vision and control stack for autonomous row-following in small student-managed crop plots. The rover maintained stable tracking across mixed lighting conditions and reduced manual guidance time during routine field observations.

This study benchmarks low-cost accelerometer nodes configured for classroom seismic alert prototypes. The best setup balanced sensitivity and false-alert control while preserving low-latency messaging performance.

The authors evaluate three compact solar charging kits designed for student use in outdoor campus areas. Over twelve clear-sky sessions, they compared charging rate, thermal behavior, and conversion efficiency under controlled load conditions. The best-performing kit delivered 18% higher average charging power than the baseline product and maintained safer operating temperatures through passive heat sinking. The paper provides practical recommendations for schools exploring resilient, low-cost renewable charging infrastructure.

This study evaluates a lightweight edge-inference workflow for near-real-time smoke-density nowcasting from low-power camera nodes. The authors compare quantized CNN variants under variable haze and lighting conditions and report latency, power draw, and forecast consistency across 600 annotated windows. The selected model reduced median inference latency by 31% while preserving prediction agreement within 4.2% of a larger cloud baseline. The paper outlines deployment trade-offs for constrained environmental sensing systems.

The author developed floating sensor pods to map pH variation along short river sections used in school field studies. Repeated deployments showed stable calibration and revealed consistent local gradients near drainage inlets.

The authors present an edge-acoustic pipeline for identifying leak-like signatures in campus utility lines using low-power microphone nodes. The approach improved early event detection while keeping battery consumption within practical maintenance windows.

The author reports temperature, pressure, and gas-flow behavior for a compact electrolyzer used in supervised student labs. The resulting safety envelope supports repeatable operation guidance for introductory clean-energy demonstrations.

The study compares linear and tree-based forecasting models for 15-minute campus microgrid demand prediction. The selected model reduced forecast error during peak transitions and supported safer battery dispatch planning.

This paper tests motion-profile tuning and software compensation methods for reducing ringing artifacts in open-frame gantry printers. A simple calibration routine lowered edge ghosting and improved dimensional consistency in repeated benchmark prints.

This work evaluates a low-cost wearable EMG band for classifying hand gestures in student assistive-interface projects. A compact classifier achieved robust performance on five gestures with minimal calibration time.

The paper presents a wavelength and intensity calibration workflow for a student-assembled open-source spectrometer. The procedure improved repeatability across sessions and reduced cross-device measurement drift.

This study compares wick materials for passive solar stills used in educational desalination demonstrations. A cotton-cellulose blend improved distillate output while maintaining stable long-run capillary transport.

This paper evaluates compressive strength and water absorption of walkway bricks produced with recycled asphalt fractions. A mid-range blend achieved favorable strength-retention trade-offs for low-load pedestrian use.

This project compares four layered filter column designs for rainwater harvested from school rooftops. Student researchers measured turbidity, conductivity, and flow rate across 40 test runs using locally available materials. A mixed-media column with gravel, fine sand, and activated charcoal achieved the best balance between clarity improvement and sustainable flow, reducing turbidity by 71% while maintaining practical throughput. The study offers a replicable protocol for classroom-level water treatment investigations.

The authors benchmark compact image-classification models for early leaf-disease screening in school greenhouse setups. The best model improved macro-F1 under class imbalance while remaining deployable on low-cost devices.

The authors benchmark reactive and map-based navigation stacks on a standardized hallway course for student robotics teams. The selected configuration reduced collision events while preserving traversal speed under clutter.

This work measures reverberation-time reduction from low-cost acoustic panel designs assembled by students. Results identify panel spacing and placement patterns that deliver the largest speech-intelligibility improvements.

This paper investigates low-complexity visual markers for autonomous drone landing exercises in introductory robotics courses. The author tested marker size, contrast, and border thickness using a simulated camera pipeline and indoor quadcopter trials. A high-contrast concentric-square marker reduced average landing error by 27% compared with common QR-pattern references while requiring less processing time. The method is intended for beginner-friendly robotics labs with limited hardware resources.

The authors present a thermal-imaging protocol for identifying early anomaly signatures in small lithium-ion battery modules. By combining frame differencing, region-threshold tracking, and cycle-level normalization, the workflow flags localized hotspots before threshold breach events. Across 120 controlled charge-discharge cycles, the method detected precursor anomalies an average of 14 minutes earlier than a voltage-only baseline. The study offers a reproducible screening framework for battery safety diagnostics.

The paper presents a compact packed-column setup for demonstrating gas absorption fundamentals with safe instructional materials. Students observed clear concentration changes and pressure-drop trends suitable for course labs.

This study evaluates a Bluetooth mesh network for synchronizing dynamic evacuation signage in school buildings. The system sustained robust update delivery during node failures and improved route guidance consistency.

This work explores starch-additive ratios for producing biodegradable foam inserts with controlled density. A narrow formulation window improved cushioning efficiency without major moisture stability losses.

This paper benchmarks a federated learning setup for acoustic event classification where raw recordings remain local to collection nodes. The authors evaluate aggregation rounds, client-dropout tolerance, and class-balance strategies using a shared protocol across five independently curated datasets. The best configuration improved macro-F1 by 9.6 points over a non-adaptive baseline while preserving data locality. Results highlight practical controls for robust privacy-preserving environmental audio analytics.

This paper proposes reproducible wind-tunnel protocols for measuring lift and drag in classroom airfoil activities. The protocol improved run-to-run consistency and reduced setup time for student teams.

The author evaluates low-cost catalyst candidates in a small hydrogen peroxide fuel-cell teaching platform. The screening approach identified stable high-activity options suitable for repeatable classroom experiments.

This paper describes a microcontroller-based irrigation controller for a mini greenhouse used in student agriculture experiments. The system uses threshold rules from soil moisture sensors and ambient temperature readings to schedule pump operation. Over a 30-day basil growth trial, automated control reduced water use by 21% compared with manual watering while maintaining similar plant height and leaf count. The implementation is designed for reproducible deployment in school laboratories.

The authors produced starch-based bioplastic films with varying glycerol concentrations and evaluated their tensile behavior for low-load packaging applications. Using a benchtop tensile rig, they measured stress-strain response, elongation at break, and humidity sensitivity. One formulation improved elongation by 34% over the control while maintaining comparable tensile strength. The paper discusses trade-offs between flexibility and moisture resistance in student-manufactured biodegradable materials.

The author models classroom thermal comfort from temperature, humidity, and airflow measurements captured by student-installed sensors. The resulting predictor aligned well with occupant surveys and supported practical ventilation adjustments.

This paper presents a baseline design for a student-operated campus energy dashboard integrating smart-meter streams, occupancy schedules, and weather context. The dashboard supported weekly audit reviews and highlighted avoidable peak-load periods in common study spaces.

The authors evaluate a lightweight simultaneous localization and mapping workflow adapted for low-cost educational robots. The pipeline produced stable maps in classroom corridors and reduced navigation drift during repeated autonomous runs.

This work deploys a distributed weather-station network across school grounds to characterize microclimate variation. The collected data exposed persistent temperature and humidity gradients between shaded and exposed locations.

The paper audits classroom network latency behavior when many student IoT devices publish telemetry simultaneously. Results identify queueing thresholds and practical routing adjustments that reduce packet delay variation.

The author studies geometric dewarping of photographed shorthand pages captured with handheld devices. The proposed correction pipeline improved line alignment and boosted downstream OCR stability.

The study evaluates non-neural denoising filters for suppressing bleed-through in scanned shorthand notes. Combined frequency and morphology filtering improved text legibility while preserving stroke continuity.

Shorthand, also known as pen stenography, is a family of writing systems for English and other languages that emerged out of a need for a fast and efficient writing system in a predigital age. Of the many English shorthand systems, Gregg shorthand is the most prevalent (Zhai et al., 2018). While largely made obsolete by general-purpose computers, the cultural and legal value within old shorthand documents means that being able to efficiently scan shorthand documents into modern computer systems holds significant value. This investigation explored the implementation of a model built around a Gated Convolutional Neural network for purposes of handwritten text recognition of Gregg shorthand. An accuracy of 40.04% was achieved after minimal training. The finalized model is freely licensed and made available online for public access.

This paper evaluates few-shot adaptation for handling writer-specific variation in Gregg shorthand recognition. Lightweight fine-tuning steps improved word accuracy for unseen writers with limited labeled samples.

The study reports inter-rater agreement metrics for a custom shorthand transcription annotation interface. Results identify annotation rules that improve consistency and reduce ambiguous symbol labeling.

This paper benchmarks projection-based and contour-based line segmentation methods on degraded cursive pages. A hybrid approach reduced merge and split errors for downstream handwriting recognition workflows.

This paper compares adaptive thresholding strategies for low-light shorthand note images captured in archival conditions. A contrast-aware pipeline improved stroke retention while reducing background artifacts.

The author evaluates rule-based and learned approaches for recovering table structure from ledger-like shorthand documents. The combined method improved row-column reconstruction and downstream field extraction reliability.

This work presents a symbol-sequence language model baseline to support shorthand transcription correction. The model reduced character-sequence perplexity and improved post-recognition consistency on held-out notes.