The project investigated the use of electric kick scooters (e‑KF) in two German cities, Berlin and Dresden, by combining automated video‑based traffic observation with a conflict‑analysis algorithm. AIT, the Austrian Institute of Technology, led the development of the algorithm, while the German Federal Office for Roads (BASt) provided funding and coordinated the field deployment. Rental operators LIME, TIER and VOI supplied anonymised usage data that were used to contextualise the observed traffic volumes. The observation campaigns were carried out from September to November 2021 at five strategically chosen sites in each city, with cameras mounted on traffic lights and lampposts at heights between 3.8 m and 4.2 m.

The algorithm was trained on a dataset of more than 100 recorded scenes, covering 26 distinct traffic situations, 25 of which involved e‑KF. This training set was deemed sufficient to achieve robust performance in the subsequent automated extraction of vehicle trajectories, speeds, and conflict indicators. The core of the conflict detection relies on two kinematic metrics: Time‑to‑Collision (TTC) and Post‑Encroachment‑Time (PET). A TTC or PET of 1.5 s or less was used to flag a potential conflict, with sub‑categories defined as critical (≤ 0.5 s), medium (0.5–1.0 s) and light (1.0–1.5 s). Only two‑party interactions were considered, as the indicators cannot capture single‑party incidents or multi‑party clashes.

In Berlin, 6 316 e‑KF were observed, representing 0.95 % of all traffic during the study period. Of these, 43 % travelled on legally permitted lanes while 57 % used pedestrian or non‑designated road space. The median speed of the e‑KF was 16.4 km/h, comfortably below the 20 km/h limit prescribed for e‑KF. Conflict analysis identified 230 incidents involving e‑KF and other road users. Eighty‑eight percent of these involved pedestrians, 7 % cyclists, and 14 % other e‑KF. The spatial distribution of conflicts showed that 74 % occurred within 1.5 m of a pedestrian, underscoring the proximity of e‑KF to foot traffic. Conflict severity was distributed as 8 % critical, 42 % medium and 50 % light. The highest rates of incorrect lane usage were recorded at Hardenbergplatz (96 %) and Brunnenstraße (83 %), whereas sites with dedicated cycling infrastructure such as Karl‑Liebknecht‑Straße (40 %) and Hannoversche Straße (28 %) exhibited lower misuse.

Dresden’s observations, conducted under similar methodological conditions, revealed a lower overall e‑KF presence. The city’s median e‑KF speed also fell below the legal limit, and the majority of vehicles adhered to designated lanes. However, due to the limited number of recorded vehicles (approximately five per hour per site) and the restriction of video analysis to daylight hours, the Dresden data set was less comprehensive. Nonetheless, the patterns mirrored those in Berlin: a tendency for e‑KF to occupy pedestrian spaces in areas lacking cycling infrastructure, and a predominance of conflicts involving pedestrians.

The project’s findings highlight that e‑KF frequently violate lane regulations, especially in pedestrian‑dense environments, and that near‑collision incidents, while relatively rare, are predominantly with pedestrians. The algorithm’s ability to automatically quantify TTC and PET from video footage provides a scalable tool for monitoring e‑KF safety and informing urban traffic management. The collaboration between AIT, BASt, and the rental operators, supported by the BASt funding, enabled a systematic assessment of e‑KF behaviour across two major German cities within a single observation season.