Data-driven bicycle network planning for demand and safety.

Developing safe infrastructure for cycling and micromobility is an efficient pathway towards climate-friendly, sustainable, and livable cities. However, urban cycling infrastructure is typically planned ad-hoc and at best informed by survey data. For a systematic, data-driven planning process here we develop an automated planning framework using data of the existing network and of empirical e-scooter trips and bicycle crashes as proxies for demand and safety, to generate a cohesive network optimizing public health benefits while minimizing investments. We introduce a parameter that tunes the focus between demand-based and safety-based development, and investigate systematically this tradeoff for the city of Turin. We find that a full focus on demand or safety generates different network extensions in the short term, with an optimal tradeoff in-between. In the long term our framework improves overall network quality independent of short-term focus. Thus, this data-driven process can provide urban planners with automated assistance for variable short-term scenario planning while maintaining the long-term goal of a sustainable, city-spanning micromobility network.