Team develops AI framework to optimize e-scooter hubs in 29 cities
AI agents now plan optimal e-scooter hub locations using causal discovery across 29 German cities.
Researchers from institutions including the Technical University of Munich and University of Stuttgart have developed an **agentic AI framework** that leverages **causal discovery** to revolutionize e-scooter mobility hub planning. Published as *arXiv:2606.25484*, the system analyzes public **GBFS data** across 29 German cities (covering 57 city-cluster units) to identify how environmental features causally drive demand for e-scooter hubs.
The framework uses an LLM-orchestrated pipeline to adapt algorithm selection to local data conditions, revealing that core demand is driven by activity access and transit proximity, while peripheral demand responds to built form. The team built a planning tool that scores candidate sites, calibrates recommendations to local demographics, and generates actionable reports. In Heilbronn, two hubs designed using this framework are now under construction, demonstrating real-world impact.
- Agentic AI framework uses causal discovery to identify city-type-specific drivers of e-scooter demand across 29 German cities
- LLM-orchestrated pipeline adapts to local data conditions and generates practitioner-ready reports for hub siting
- Two real-world hubs in Heilbronn, Germany, are under construction based on framework outputs
Why It Matters
Moves urban mobility planning from correlation-based guesswork to causal, data-driven decision-making for scalable e-scooter infrastructure.