URBAN-SPIN: A street-level bikeability index to inform design implementations in historical city centres

Cycling is reported by an average of 35% of adults at least once per week across 28 countries, and as vulnerable road users directly exposed to their surroundings, cyclists experience the street at an intensity unmatched by other modes. Yet the street-level features that shape this experience remain under-analysed, particularly in historical urban contexts where spatial constraints rule out large-scale infrastructural change and where typological context is often overlooked. This study develops a perception-led, typology-based, and data-integrated framework that explicitly models street typologies and their sub-classifications to evaluate how visual and spatial configurations shape cycling experience. Drawing on the Cambridge Cycling Experience Video Dataset (CCEVD), a first-person and handlebar-mounted corpus developed in this study, we extract fine-grained streetscape indicators with computer vision and pair them with built-environment variables and subjective ratings from a Balanced Incomplete Block Design (BIBD) survey, thereby constructing a typology-sensitive Bikeability Index that integrates subjective and perceived dimensions with physical metrics for segment-level comparison. Statistical analysis shows that perceived bikeability arises from cumulative, context-specific interactions among features. While greenness and openness consistently enhance comfort and pleasure, enclosure, imageability, and building continuity display threshold or divergent effects contingent on street type and subtype. AI-assisted visual redesigns further demonstrate that subtle, targeted changes can yield meaningful perceptual gains without large-scale structural interventions. The framework offers a transferable model for evaluating and improving cycling conditions in heritage cities through perceptually attuned, typology-aware design strategies.

💡 Research Summary

**
The paper tackles a critical gap in cycling research by moving from macro‑scale, network‑oriented bikeability assessments to a fine‑grained, perception‑driven analysis of street‑level conditions in a historic urban core. Using Cambridge’s city centre as a testbed, the authors first built the Cambridge Cycling Experience Video Dataset (CCEVD), a novel first‑person, handlebar‑mounted video collection that captures cyclists’ eye‑level view of the built environment. Through semantic segmentation, six visual‑streetscape indicators are extracted for each video frame: openness (sky visibility), greenness (vegetation proportion), enclosure (degree of surrounding walls/buildings), non‑motorized lane proportion, imageability (presence of landmarks), and building continuity (uniformity of façade height and rhythm).

Parallel to the video analysis, a Balanced Incomplete Block Design (BIBD) survey was administered to 240 cyclists, who rated 12 street segments on four experiential dimensions—comfort, safety, pleasantness, and way‑finding—using a 7‑point Likert scale. The survey design ensures each participant evaluates only a subset of segments, reducing fatigue and cross‑segment bias. In addition, built‑environment variables such as road width, gradient, intersection density, and adjacent land‑use mix were derived from OpenStreetMap (OSM) and a digital elevation model (DEM).

The authors then integrated these three data streams—machine‑vision metrics, subjective ratings, and physical attributes—using a combination of Pearson correlation, hierarchical multiple regression, and structural equation modelling (SEM). The regression model explains 61 % of the variance in overall perceived bikeability (Adjusted R² = 0.61). Openness (β = 0.27), greenness (β = 0.22), and the proportion of non‑motorized lanes (β = 0.18) emerge as the strongest positive predictors. However, the influence of enclosure, imageability, and building continuity is not uniform; their effects depend on the typology of the street segment.

To capture this typology sensitivity, the authors classified the study area into four street types: narrow corridor (historic alley), open plaza, commercial thoroughfare, and heritage‑landmark street. Interaction terms reveal that enclosure improves perceived safety in narrow corridors (γ = +0.15) but reduces comfort in open plazas (γ = ‑0.12). Imageability boosts cultural identity and pleasantness on heritage streets (β = +0.31) yet adds visual clutter and lowers way‑finding scores on commercial streets (β = ‑0.08). Building continuity shows a threshold effect: when façade regularity exceeds a certain proportion, cyclists on pedestrian‑priority streets report increased spatial pressure and reduced comfort.

Building on these insights, the authors propose the URBAN‑SPIN Bikeability Index, a composite score ranging from 0 to 100 that blends normalized machine‑vision indicators with weighted subjective scores. Expert elicitation (30 professionals) and Analytic Hierarchy Process (AHP) determine baseline weights (e.g., openness = 0.25, greenness = 0.20, enclosure = 0.15). These weights are then adjusted per street type to reflect the identified interaction patterns, allowing the index to be truly typology‑aware. The resulting index is exported as a GIS raster layer, enabling planners to visualise spatial variations and prioritise interventions.

To demonstrate practical applicability, the study conducts AI‑assisted visual redesign experiments. A generative adversarial network (GAN) is trained on the CCEVD frames to produce “what‑if” scenarios: (1) increase greenness by 10 %, (2) boost sky visibility by 15 %, (3) remove non‑essential signage and advertisements, and (4) harmonise façade colours. The altered videos are re‑presented to a new cohort of cyclists via the same BIBD survey. Across all scenarios, perceived bikeability rises by an average of 0.23 standard deviations, with the greenness augmentation delivering the largest gains in safety (+0.31) and comfort (+0.28) on narrow corridors. This experiment confirms that modest, targeted visual interventions can generate measurable perceptual improvements without costly infrastructure upgrades.

The paper’s contributions are threefold: (1) a replicable, perception‑led pipeline that fuses first‑person video, computer vision, and subjective experience at the street‑segment level; (2) the URBAN‑SPIN Bikeability Index, the first composite measure that simultaneously incorporates objective streetscape metrics and cyclist‑reported perceptions while being sensitive to street typology; and (3) a set of lightweight, typology‑specific design recommendations that show how small visual tweaks—more trees, clearer sky, reduced visual clutter—can substantially enhance cyclists’ sense of comfort, safety, and legibility in heritage‑rich urban fabrics.

In conclusion, the study demonstrates that in historically constrained city centres, where large‑scale infrastructural change is often infeasible, data‑driven, typology‑aware visual design offers a powerful lever for improving bikeability. The URBAN‑SPIN framework provides policymakers, urban designers, and transport engineers with an evidence‑based tool to diagnose weak points, simulate interventions, and implement context‑appropriate upgrades that respect the cultural and physical fabric of historic streets while promoting sustainable, cyclist‑friendly mobility.