Quantify, diagnose, and map 10-minute pedestrian reach across urban street networks at native CPU speed.
walkability computes the reachable walkable network length within a 10-minute walk (~750 m at 4.5 km/h) for every junction in a city. Designed for urban planners and researchers, it analyzes OpenStreetMap street data or user-provided road networks with a multi-threaded Rust kernel and exact partial-edge credits.

Reachable network length in Cusco. Yellow highlights dense, highly permeable pedestrian districts (Historic Center); dark purple marks fragmented outskirts and steep hillsides.
Quickstart
Installation
Install the development version from GitHub:
# install.packages("pak")
pak::pak("PaulESantos/walkability")Requirements for source installations
R 4.2 or greater is required.
A working Rust toolchain is required when installing walkability from source. Install Rust from rust-lang.org/tools/install. The additional requirements are:
- Cargo 1.80 or greater, the Rust package manager.
- Rustc 1.80 or greater, the Rust compiler.
On Windows, install the matching version of Rtools first. Then, from a terminal in the package project directory, the most practical Rust target for an R + Rtools source build is:
CRAN binary installations on Windows and macOS do not normally require a Rust installation; this requirement applies when compiling the package from source.
Reproducible Example with Real Urban Data
walkability includes real, curated street network datasets from OpenStreetMap for Cusco, Lima, and Arequipa:
library(walkability)
library(sf)
# 1. Load real street network from Cusco (Historic Center, UTM 19S)
data(streets_cusco)
# 2. Build graph network & compute pedestrian reach (Rust)
net <- walk_build_network(streets_cusco, walk_speed_kmh = 4.5, verbose = FALSE)
reach <- walk_compute_reach(net, time_min = 10, verbose = FALSE)
# 3. View results
reach── Walkability Reachable Network Length ────────────────────────────────────────
✔ Isochrone Budget: 10.0 minutes (~750 m at 4.5 km/h)
✔ Evaluated Nodes: 249
• Average Reach: 8.75 km
• Highest Reach: 17.39 km
• Lowest Reach: 0.75 km
• 5th-95th percentile: 3.1 - 15.5 km
# 4. Point-to-point pedestrian routing & circuity
orig <- st_coordinates(net$nodes[net$nodes$is_giant, ])[1, 1:2]
dest <- st_coordinates(net$nodes[net$nodes$is_giant, ])[50, 1:2]
path <- walk_shortest_path(net, from = orig, to = dest)
path
# 5. Isochrone polygon generation (concave hull)
iso <- walk_isochrone(net, origin = orig, time_min = c(5, 10))
iso
# 6. Urban intersection density & morphology
dens <- walk_intersection_density(net)
densKey Capabilities
-
Curated Real-World Datasets: Out-of-the-box model networks for
streets_cusco,streets_lima, andstreets_arequipa. -
Non-destructive Topological Diagnosis:
walk_diagnose_network()inspects user street files, flags isolated islands, and locates near-miss undershoots without mutating geometries. -
Controlled Geometric Repair:
walk_build_network()empowers users to choose whether to planarize crossings and snap dangling endpoints at specified tolerances. -
Exact Partial-Edge Dijkstra in Rust:
walk_compute_reach()computes exact fractional contributions for boundary edges across all CPU cores with zero heap reallocations. -
Pedestrian Point-to-Point Routing:
walk_shortest_path()extracts the exact shortest path geometry, travel time, and route circuity detour ratio. -
Isochrone Polygons & Subgraphs:
walk_isochrone()generates exact reachable network corridors and concave hull polygons. -
Cumulative POI Accessibility:
walk_compute_access()counts reachable amenities (schools, healthcare, transit) within travel time budgets using parallel Rust. -
Urban Morphology & Intersection Metrics:
walk_intersection_density()classifies cul-de-sacs, 3-way T-junctions, 4-way crossings, and intersection density. -
Intervention Scenario Simulation:
walk_scenario_close()andwalk_compare_scenarios()model street cuts, closures, and pedestrianization. -
Graph Interoperability:
as_sfnetwork()andas_tidygraph()enable two-way workflows withsfnetworksandtidygraph. -
Publication-Ready Maps:
walk_plot_reach()renders cleanggplot2visualizations with 2nd–98th percentile scaling and ascending layer stacking.
Data Sources and Attribution
The bundled Cusco, Lima, and Arequipa street networks are derived from OpenStreetMap data and are available under the Open Database License (ODbL). When using, redistributing, or adapting these datasets, retain the attribution © OpenStreetMap contributors and comply with the ODbL.
You may also analyze a custom street network supplied as an sf object or uploaded through the application. You are responsible for confirming that you have the right to use and redistribute any user-provided data.
Scope & Limitations
- Modeled speed: Assumes a design walking speed of 4.5 km/h (1.25 m/s). It measures structural street connectivity, not observed GPS pedestrian speeds.
- Footway availability: All non-motorway and non-trunk roads are modeled as walkable unless tagged as closed to pedestrians.
- Planar topology: Grade-separated intersections (overpasses) are treated as planar unless explicit elevation tags exist.
Read the complete Methodology and Model Assumptions guide for details.
Documentation & Vignettes
- Guía de Inicio Rápido: Complete step-by-step workflow with user-provided datasets, network repair, and diagnostic plots.
- Metodología, Algoritmo Dijkstra en Rust y Supuestos: Mathematical derivation of the partial-edge rule, CSR representation, and 11 modeling limitations.
- Resultados de Referencia en Ciudades: Benchmark findings and verified network statistics for Cusco, Arequipa, and Lima.
Citation & License
To cite walkability in publications, run citation("walkability"). The current software citation is:
Santos Andrade, P. E. (2026). walkability: Urban Walkability and Active Transit
Network Analysis. R package version 0.1.0.
https://github.com/PaulESantos/walkability
Distributed under the MIT License.