This project adapts the School Vehicle Arrival Departure (SVAD) algorithm into a general-purpose geospatial tool for urban contexts. The core idea: transform raw GPS breadcrumbs into auditable arrival, departure, and dwell events at Points of Interest (POIs) such as schools, depots, transit stops, hospitals, or loading docks.

• Deliverables: Literature review, architecture outline

• Activities:

  • Review SVAD design and related geospatial algorithms (map-matching, stop detection, H3/S2 spatial indexing).
  • Survey urban use cases: public transit, freight, curb management, emergency response.
  • Define project goals and scope (what events, what POIs, what data streams).
  • Document evaluation metrics (precision/recall of arrival detection, dwell time accuracy).

• Deliverables: Clean sample dataset and POI registry

• Activities:

  • Collect GPS breadcrumb datasets (from open transit feeds, sample telematics, or simulators).
  • Compile a small POI dataset (schools, depots, loading zones) with polygons and entrance cells.
  • Normalize coordinate systems, ensure timestamps are synchronized.
  • Explore map-matching libraries to snap breadcrumbs to street networks.

• Deliverables: Stop detection prototype

• Activities:

  • Implement stop detection logic (speed < threshold, gear=park if available, stopped > N seconds).
  • Handle noise, short pauses, and urban canyon GPS drift.
  • Validate on sample data: mark detected stops vs. raw GPS traces.
  • Document false positives (e.g., traffic lights) and brainstorm filters.

• Deliverables: Event log generator (arrival, departure, dwell)

• Activities:

  • Use spatial indexing (H3/S2) to quickly map stops to POIs.
  • Add entrance-cell logic to disambiguate curbside dwell from true arrival.
  • Generate structured events: {arrival, departure, dwell duration, POI ID, confidence}.
  • Run QA: compare auto-detected events against ground-truth timetables (if available).

• Deliverables: Prototype dashboard & metrics report

• Activities:

  • Aggregate event logs into KPIs: on-time %, average dwell, missed arrivals.
  • Create simple visualizations (heatmaps, timelines, Gantt-style fleet charts).
  • Compare across POI types: schools vs. depots vs. loading docks.
  • Evaluate computational efficiency and scalability.

• Deliverables: Final report, GitHub repo, demo notebook

• Activities:

  • Document system architecture, data pipeline, algorithms.
  • Write reproducible Jupyter notebooks for data prep, stop detection, and event generation.
  • Identify limitations (GPS noise, POI definition quality, privacy considerations).
  • Propose future extensions: streaming pipeline, privacy-preserving aggregation, ML-based stop classification.

• A working prototype of a general-purpose SVAD engine.
• Demonstrated use on at least one urban dataset (transit, freight, or public service).
• Clear metrics on detection accuracy and performance.
• Open-source repo with documentation for reuse and extension.

The generalized SVAD tool could provide value across multiple urban systems:

• Public Transit: Automatic bus arrival/departure detection to improve on-time performance reporting and headway management.
• Freight & Logistics: Tracking arrival and dwell at loading docks or distribution centers for better curbside management and dynamic scheduling.
• Emergency Services: Logging ambulance or firetruck arrivals at scenes and hospitals for response time analytics.
• Micromobility: Identifying scooter and bike arrivals at parking zones to enforce compliance and rebalance fleets.
• Urban Operations: Detecting when waste trucks, snow plows, or street sweepers actually serviced blocks.
• Healthcare & Social Services: Measuring arrival and service dwell for mobile clinics or food trucks to evaluate coverage.
• Infrastructure & Utilities: Logging crew arrival and dwell at work sites for SLA monitoring.