A) Why trucks?
- Coverage: every neighborhood, curb‑level sampling, repeatable weekly transects.
- Co‑benefits: the same platform supports safety, operations, and environmental research.
- Equity lens: pair hyperlocal exposures with demographics/land use to study environmental justice.
B) Research themes (examples)
| Theme | What to measure | Questions you can answer | Refs |
|---|---|---|---|
| Air‑quality exposure | PM2.5/PM10, NO2, O3, CO, BC; GNSS/time | Hotspots near schools/arterials; diurnal/seasonal patterns; impacts of traffic/policy changes | NASA trash‑truck AQ case • Apte et al. 2017 (PNAS) • City Scanner 2023 |
| Methane & odors | CH4 (NDIR/laser), TVOC (PID/MOS) | Leak screening; odor indices near transfer stations/food‑waste routes; mitigation evaluation | Google/EDF methane mapping • PID guide |
| Noise & soundscapes | dBA/octave logging, events (sirens, horns) | Who bears highest noise; effects of route time shifts; link to well‑being | City Scanner 2023 |
| Urban heat & microclimate | Air temp/RH, globe temp, IR surface temp | Block‑level heat patterns; cooling benefits (trees, reflective pavement); schedule changes | Notre Dame UHI 2025 |
| Waste behavior & contamination | Hopper camera CV, on‑board scales, RFID | Overflow frequency; contamination spatial patterns; education or pricing impacts | 3rd Eye CV • Compology coverage |
| Battery fire risk | Thermal cam, Temp/CO/TVOC combo | Early off‑gas cues; incident origin analysis; policy for LIB disposal | US EPA LIB fires |
| Road dust & infrastructure | PM spikes + vibration/IRI proxy | Resuspension risk; prioritize street maintenance for health co‑benefits | Road roughness via accelerometers (2024) |
C) Study designs & methods
- Before/after, difference‑in‑differences around interventions (containerization, time‑of‑day changes).
- Mobile monitoring best practices: collocate with reference monitors; compensate for T/RH; remove self‑heating bias; aggregate to street segments.
- Models: Land‑Use Regression (LUR), kriging, and Bayesian hierarchical models for spatiotemporal fields.
- Uncertainty: bootstrap confidence on segment medians; track sensor drift and calibration windows.
D) Data architecture
| Layer | Common tools | Notes | Refs |
|---|---|---|---|
| Edge | Pi 5/Jetson; PM/gas/BC/noise; GNSS; event cameras | Event‑triggered clips; encrypt at rest; version sensors & models | — |
| Backhaul | MQTT/HTTPS; buffer & retry | Publish JSON records per segment/event | MQTT |
| Storage/analysis | PostGIS; notebooks (R/Python) | Segment join, QA/QC, models | PostGIS |
| GIS & dashboards | ArcGIS / QGIS; Dashboards/Lizmap | Temporal playback; EJ overlays; story maps | ArcGIS Dashboards • Lizmap |
| Policy integration | CalEnviroScreen/EJScreen; open data portals | Join exposures with EJ indicators | CalEnviroScreen • EPA EJScreen |
E) Governance & ethics
- Minimize PII: avoid plates/faces; if captured, blur at the edge; publish derived layers (rasters/segments), not raw imagery.
- Transparency: document what’s collected, purpose, retention, and calibration—plain language summaries for the public.
- Quality: collocation checks; periodic calibration; data dictionaries; public uncertainty notes.
F) Teaching & community engagement
- Course modules: Mobile air & EJ, Urban heat transects, Waste behavior & contamination, Noise & well‑being.
- Participatory science: co‑design routes with community boards; validate hotspots with stationary sensors or school monitors.
- Deliverables: segment rasters, EJ overlays, one‑page policy briefs per finding.
G) Quick‑start kit (illustrative)
| # | Component | Purpose | Notes |
|---|---|---|---|
| 1 | PM sensor (SPS30/OPC‑N3) | PM1/2.5/10 | Shielded intake; humidity logging |
| 2 | Gas sensors (NO2, O3, CO, CO2) | Exposure indices | Electrochemical + NDIR |
| 3 | TVOC (PID/MOS) | Odor/solvent proxy | PID does not detect CH4 |
| 4 | Methane (NDIR/laser) | Leak screening | Optional research‑grade TILDAS |
| 5 | Black Carbon (microAeth) | Diesel soot | Advanced option |
| 6 | Noise mic (Class 2) | Noise mapping | Wind screens; calibration |
| 7 | GNSS + IMU | Geo/time alignment | Optionally RTK for campaigns |
| 8 | Edge compute (Pi 5/Jetson) | Buffering; QC; analytics | Containerize services |
H) Example data schema (JSON)
{
"truck_id": "TRUCK-12",
"segment_id": "SEG-2025-10-05-0142",
"time_utc": "2025-10-05T14:02:00Z",
"gps": {"lat": 37.7812, "lon": -122.4058, "speed_mps": 7.5},
"sensors": {
"pm25_ugm3": 18.4, "pm10_ugm3": 30.1,
"no2_ppb": 21, "co_ppm": 0.3, "tvoc_ppb": 120,
"bc_ugm3": 1.2, "noise_dba": 68.3, "air_temp_c": 28.1, "rh_pct": 46
},
"flags": ["school_zone","hot_day"],
"qa": {"pm25_valid": true, "bc_qc": "ok", "cal_version": "2025-09-03"}
}Sources (selected)
- Mobile trucks for AQ/urban sensing: NASA/deSouza 2020 • MIT City Scanner 2023 • Apte et al. 2017 (PNAS)
- Methane & VOC methods: Google/EDF methane mapping • Photoionization detector guide
- Noise & heat examples: City Scanner • Notre Dame UHI (2025)
- EJ indicators: CalEnviroScreen • EPA EJScreen
- Road/IRI link: NDSU 2024 report