← Back to selected projects

03 / Ongoing research · Safe mobility

Geosocial Risk Modeling

An explainable spatial-AI project exploring how the environment around railroad crossings and broader community patterns can support proactive trespassing-risk prioritization.

  • Spatial AI
  • Risk Modeling
  • Explainable AI
  • Railroad Safety
Conceptual geosocial risk landscape surrounding an urban railroad grade crossing
Conceptual risk landscapePublic-facing visualization

Challenge

Move from incident response to proactive safety planning.

Railroad trespassing risk is influenced by more than a crossing alone. The surrounding place, the community it serves, and its connection to nearby locations all shape the local safety context.

Research goal

Identify priority locations and make the reasoning understandable.

The project investigates an interpretable way to estimate relative risk while showing the broad contextual evidence behind each assessment.

Public research scope / 01

Three complementary views of place

01

Physical context

The built and geographic environment surrounding a railroad crossing.

02

Community context

Aggregated characteristics that describe the communities around each location.

03

Regional context

Relationships among nearby crossings and the wider transportation landscape.

04

Interpretable evidence

High-level explanations designed to connect a risk estimate with actionable local context.

Intended value / 02

Decision support for safer communities

  1. 01

    Prioritize

    Help transportation stakeholders identify locations that may warrant closer attention.

  2. 02

    Understand

    Provide interpretable context instead of presenting risk estimates as a black box.

  3. 03

    Act

    Support targeted monitoring, education, planning, and safety-resource decisions.

Current stage / 03

Research in progress

Status

Model development and validation are ongoing.

The framework is being refined as part of a manuscript in preparation. This page intentionally presents only the public research direction.

Publication boundary

Technical details will follow peer review.

Architecture, variables, experimental settings, validation procedures, and results are withheld until the work is ready for publication.