Overview

The Safe Streets Planner provides a comprehensive solution for assessing and prioritizing high-risk corridors and generating detailed reports.

This application combines our detailed multimodal data and driving event data from Michelin Mobility Intelligence, enabling you to map existing conditions, analyze specific corridors, and generate safety action plans and reports. By fusing metrics like speeding, hard-braking, and distracted driving with data on mobility activity and demographic information, this application empowers agencies to identify, prioritize, intervene, and monitor impacts on high conflict corridors in their communities.

Key Features:

• Customize parameters for high-risk corridors.
• Rank corridors by risk percentage.
• Analyze driver events and assess vulnerable road user information per corridor.
• Evaluate the effectiveness of prioritized corridors in resolving conflicts.
• Communicate and collaborate with project team members directly within the application
• Create an action plan for sharing with stakeholders and integrating into project deliverables.

Safe Streets Planner in Action

The application includes the following components:

• Explore View: Customize risk score parameters to visualize, explore, and prioritize corridors. See a ranked list of corridors by risk.

• Corridor-Focused View: Select individual corridors from the map to see a comprehensive breakdown of their risk score components, pinpointing specific areas of concern.

• Planner View: Strategically prioritize high-risk corridors, collaborate with project teams, and develop action plans for sharing with stakeholders and integrating into project plans.

Methodology

Risk Scores

The Safe Streets Planner incorporates both observed driver event data and Replica's mobility and demographic to assign risk scores to corridors in a given area. A corridor's risk score is generated using the following logic:

1. For each segment or corridor, we multiply the “total driver events” (filtered) x “total trips” (filtered) = risk potential per segment or corridor
2. To normalize, we then divide each corridor’s “risk potential per corridor” by the corridor’s mileage = risk potential per corridor per mile
3. We then take rank each segment or corridor’s “risk potential per mile” by percentile = risk score
   1 = lowest risk potential, 100 = highest risk potential.

% of RISKY DRIVING = “Driver events per corridor” / “Total driver events” (100)

Driver Event Data

The driver data included in the risk score is sourced from Michelin Mobility Intelligence (MMI). They capture anonymous driver behavior from over 40 million connected drivers nationwide, including events such as:

• Suspected Collision: Detected when there is a sharp spike in accelerometer followed by a major deceleration.
• Phone Handling: Identified when using the phone’s gyroscope, accelerometer as well as unlocking function.
• Sudden Acceleration: A sudden increase in speed beyond a certain threshold for a certain time. Detected when it exceeds 7.1 mph/s.
• Sudden Breaking: A decrease in accelerometry above a certain threshold for a certain time. Detected when speed falls below -7.1 mph/s.
• Speeding Events:
  • 10mph or More Over Speed Limit: Detected when speed is 10mph or more over speed limit.
  • 20mph or More Over Speed Limit: Detected when speed is 20mph or more over speed limit.
  • Excessive Speeding (80mph+): Detected when speed exceeds 80 mph.

MMI sources its data from mobile phones. Data is collected from embedded software development kit (SDK) in consumer mobile applications that users opt into, such as Allstate, Life 360, Credit Karma, Gas Buddy, and more. At a census block level, MMI typical capture 15-25% of all licensed drivers.

For each event, data including time, location, GPS coordinates, and speed is captured and contextualized by weekday or weekend. The speed data that is captured is done via a speedometer from smartphone SDK. All driver event data is then integrated with OpenStreetMap’s (OSM) network for segment-level aggregation.

Mobility and Demographic Data

The total trips data included in the risk score is based on modeled mobility and demographic data from Replica’s seasonal (Places) dataset. This data is generated by running large-scale, computationally intensive simulations. To create Places models, Replica sources a composite of data sources and follows the below steps:

• Create a synthetic population that matches the characteristics of a given region
• Train a number of behavior models specific to that region
• Run simulations of those behavior models applied to the synthetic population in order to create a “replica” of transportation and economic patterns
• Calibrate the outputs of the model against observed “ground-truth” to improve quality

The output of this dataset includes detailed trip and population tables for a typical weekday and weekend day in a given season. Each trip is assigned a set of attributes, including start and end times, origin and destination information, primary mode, trip routing behavior, and, and trip-taker demographics.

More detailed information about the process Replica follows to create these models can be found below:
Seasonal Mobility Model Methodology (Summary)
Seasonal Mobility Model Methodology (Extended)

This data allows filtering by trips traveling through specific road segments, and is also joined to OSM’s roadway network. This facilitates seamless integration with driver event data from Michelin Mobility Intelligence (MMI).

Risk Score Filters

Set custom parameters for your risk analysis in the Filter Panel. See below for a full list of filters and their definitions: