homeresearchprojectsaboutcontact
← Projects
the problem

Despite its reputation as a global cycling capital, Amsterdam identifies dangerous infrastructure only after harm has occurred. The current system relies on accident reports and citizen complaints. Near-misses, braking events, swerves — remain invisible to planners.

Accident
Investigate
Fix
270
cyclist deaths across the Netherlands, 2023
2,507
ambulance calls in Amsterdam, 2022
+2,507
estimated serious accidents with no ambulance involved
untracked
minor incidents and near-misses entirely undocumented
approach
01
Sensor data
Bike-mounted sensors collected GPS, acceleration, and speed every second across 110 trips through Amsterdam. 718,303 total readings.
02
ABM simulation
Each trip becomes an agent navigating the Amsterdam road network. Hard braking events trigger probabilistic state transitions from SAFE to AT_RISK to UNSAFE.
03
Infrastructure overlay
Braking events mapped against traffic lights, road surface type, and facility type to isolate infrastructure-driven risk.
04
Hotspot identification
Clusters of unexpected braking away from signals point to structural failures before accidents occur.
the model

Each of the 110 recorded cycling trips is initialised as an independent agent. At every step, an agent occupies one of three states. Hard braking — defined as acceleration below -2.0g — triggers a probabilistic transition to a worse state.

SAFE
acc < -2.0gp = 0.033
AT RISK
absorbing
UNSAFE
Hard braking threshold
acc < -2.0g
Calibrated from the acceleration distribution across 718,303 sensor readings. Captures unexpected, sudden braking — a behavioural signal of infrastructure stress.
Distribution of acceleration values (n = 718,303)
-2.0g
-12g (hard braking)0g+15g
Hard braking
Normal cycling
key findings
Finding 01
55.5%
of simulated trips ended in an UNSAFE state

Unsafe trips accumulated on average 730 hard braking events compared to just 97 for safe trips. A 7x difference pointing to systemic, route-level risk.

730
unsafe avg
braking events
vs
97
safe avg
braking events
Finding 02
93.9%
of hard braking occurs away from traffic lights

Traffic signals account for only 6.1% of all hard braking. The danger is in the infrastructure between stops — surfaces, lane width, mixed-use paths.

Away from signals
93.9%
Near signals
6.1%
Finding 03
3-4x
more braking per km on brick and tile surfaces

Brick pavers produce 30 braking events per km versus 8-9.5 on asphalt. Amsterdam's historic surfaces are a structural safety hazard.

Braking events per km by road surface
Brick pavers
30 / km
Tiles
28 / km
Asphalt (red)
9.5 / km
Asphalt (gray)
8.0 / km
Finding 04
Centraal
Amsterdam Central emerges as the primary danger hotspot

Braking clusters concentrate around Amsterdam Centraal even after removing signal-adjacent events, pointing to mixed infrastructure and legacy surface materials.

Amsterdam Centraal — ABM hotspot output
team

ABM design and implementation: Marfa Kozelets
Team: M. Balancier, K. Hänni, C. Xia
UvA Computational Social Science

GitHub report on request