Overview
A real-time, browser-based discrete-event simulation of a Zipline-style drone delivery logistics system for San Francisco. The simulation features autonomous drones fulfilling delivery orders across real SF neighborhoods with live performance metrics, allowing users to watch drone animation in real-time on an interactive map.
Problem
Urban air mobility and drone delivery systems require complex logistics planning and optimization. Organizations need to understand fleet sizing requirements, service area coverage, capacity planning under varying demand, and the impact of operational parameters (drone speed, service times, etc.) on system performance without deploying actual infrastructure. Traditional approaches to logistics planning require expensive physical prototypes or complex simulation software that requires significant technical expertise to operate.
Solution
A zero-backend, single-page web application that simulates complete drone delivery operations with mathematical accuracy. The system provides:
Core Features:
- Real-time drone animation between hub and 8km delivery radius on interactive map
- Live performance metrics including average delivery time, P95 wait time, and completion rate
- Dynamic simulation controls supporting up to 1000 drones with adjustable speeds and time scaling up to 50x
- Real SF building locations sourced from OpenStreetMap for geographic accuracy
- Discrete-event simulation engine with Poisson arrivals and greedy scheduling algorithms
Use Cases:
- Fleet sizing optimization to determine optimal drone quantities for demand levels
- Service area analysis to evaluate coverage and accessibility
- Capacity planning under varying demand scenarios
- Parameter sensitivity testing for operational decision-making
The entire simulation runs in the browser with no backend infrastructure required, making it instantly accessible and deployable.
Technical Architecture
Built as a zero-dependency single-page application that runs entirely in the browser, the system demonstrates how sophisticated logistics simulations can be delivered without backend infrastructure.
Performance Optimization:
The simulation handles up to 1000 concurrent drones with real-time position updates and metric calculations. Time scaling up to 50x enables rapid scenario evaluation while maintaining simulation accuracy.
No frameworks, build tools, or backend infrastructure required - the entire application runs client-side with minimal dependencies.
Challenges
Creating a performant discrete-event simulation that could handle 1000+ concurrent drones while maintaining real-time visualization required careful optimization of the event processing loop and position update calculations. Integrating real geographic data from OpenStreetMap while keeping the application lightweight meant developing efficient data loading and caching strategies that wouldn't compromise the zero-backend architecture. Balancing simulation accuracy with visual responsiveness was crucial - the system needed to provide mathematically sound results while remaining interactive and visually engaging for users exploring different scenarios.
Results & Impact
The simulation successfully demonstrates that sophisticated logistics analysis tools can be delivered as lightweight web applications without backend infrastructure. Users can instantly explore fleet sizing scenarios, analyze service area coverage, and understand the impact of operational parameters on system performance.