# US 2025/0085135 A1 · Method and system for measuring extrinsic crash risk

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0085135 A1
- Document ID: US20250085135A1
- Status: Published application
- Kind: A1
- Application number: 18/827,185
- Filed: 2024-09-06
- Published: 2025-03-13
- Earliest claimed: 2023-09-08
- Assignees: Cambridge Mobile Telematics Inc.
- Applicants: Cambridge Mobile Telematics Inc.
- Inventors: John Collins; Jun-Geun Park; Ruilin Chen; Patrick Steele; Hari Balakrishnan; Yulan Liang
- Web page: https://patentgrep.com/patent/US20250085135A1

## CPC

- G01C 21/3804: {Creation or updating of map data}
- G01C 21/3859: {Differential updating map data}
- G06N 3/042: Knowledge-based neural networks; Logical representations of neural networks
- G08G 1/0112: {from the vehicle, e.g. floating car data [FCD]}
- G08G 1/0129: {for creating historical data or processing based on historical data}
- G08G 1/0133: {for classifying traffic situation}
- G08G 1/0141: {for traffic information dissemination}
- G08G 1/0145: {for active traffic flow control}
- G08G 1/096816: {where the complete route is transmitted to the vehicle at once}
- G08G 1/096827: {where the route is computed onboard}
- G08G 1/096838: {where the user preferences are taken into account or the user selects one route out of a plurality}
- G08G 1/096844: {where the complete route is dynamically recomputed based on new data}
- G08G 1/16: Anti-collision systems (road vehicle drive control systems for predicting or avoiding probable or impending collision otherwise than by control of a particular sub-unit B60W30/08)
- G08G 1/164: {Centralised systems, e.g. external to vehicles}
- G01C: MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY (measuring liquid level G01F; radio navigation, determining distance or velocity by use of propagation effects, e.g. Doppler effects, propagation time, of radio waves, analogous arrangements using other waves G01S)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G08G: TRAFFIC CONTROL SYSTEMS (guiding railway traffic, ensuring the safety of railway traffic B61L; arrangement of road signs or traffic signals E01F9/00; radar or analogous systems, sonar systems, lidar systems specially adapted for traffic control G01S13/91, G01S15/88, G01S17/88; {radar or analogous systems, sonar systems, lidar systems specially adapted for anti-collision purposes G01S13/93, G01S15/93, G01S17/93})

## Abstract

Techniques for measuring extrinsic crash risk are provided. In some examples, a plurality of segments are instantiated in a road segment datastore, with each segment corresponding to a physical road segment in a geographic area and including segment attributes derived from digital map data as well as a risk attribute. From the digital map data, pairs of physical road segments that intersect are determined such that each pair has a corresponding pair of segments in the road segment datastore, and each segment is included in at least one pair. A connection between each segment in each pair is created in the datastore comprising connectivity attributes derived from the digital map data. Based on the segment attributes for a segment, as well as the one or more paired segments and their connectivity attributes, the risk attribute for each segment is updated in the road segment datastore.

## Claims

### Claim 1

A computer-implemented method of mapping driving risk in a geographic area, the computer-implemented method comprising:
instantiating, in a road segment datastore, a plurality of segments, wherein each segment of the plurality of segments has a corresponding physical road segment of a plurality of physical road segments in the geographic area, and wherein each segment of the plurality of segments comprises:
segment attributes derived from digital map data for the corresponding physical road segment; and
a risk attribute;
determining, from the digital map data, pairs of physical road segments that intersect, wherein each pair of physical road segments that intersect has a corresponding pair of segments in the road segment datastore, and each segment of the plurality of segments is included in at least one pair of segments;
creating, in the road segment datastore, a connection between each segment in each pair of segments, wherein the connection between each pair of segments in the road segment datastore comprises connectivity attributes derived from the digital map data; and
updating, in the road segment datastore, the risk attribute for each segment of the plurality of segments that is connected to one or more paired segments based on:
the segment attributes for the segment;
the segment attributes for the one or more paired segments; and
the connectivity attributes between the segment and the one or more paired segments.

### Claim 2 (depends on 1)

The computer-implemented method of claim 1, wherein the connectivity attributes include one or more of a ramp direction attribute, a speed limit differential attribute, a number of segments in proximity to an intersection attribute, an intersection angle attribute, movement restriction attributes, a cross traffic attribute, or a cross traffic protection attribute.

### Claim 3 (depends on 1)

The computer-implemented method of claim 1, wherein the risk attribute for each segment of the plurality of segments is updated by a segment-level crash-risk prediction model executed on the plurality of segments and the connection between each pair of segments in the road segment datastore.

### Claim 4 (depends on 3)

The computer-implemented method of claim 3, wherein the segment-level crash-risk prediction model uses a graph neural network (GNN) and the computer-implemented 2 method further comprises:
receiving truth data comprising locations of known vehicle crashes within the geographic area; and
training the GNN to predict a crash rate for each segment using the truth data.

### Claim 5 (depends on 4)

The computer-implemented method of claim 4, wherein the GNN is trained to minimize a value of a Poisson loss function that provides an expected number of crashes for a given segment of the plurality of segments.

### Claim 6 (depends on 1)

The computer-implemented method of claim 1, wherein the risk attribute for a segment represents a crash rate per traversal of the segment and the computer-implemented method further comprises:
receiving, from a mobile device, a start location and an end location within the geographic area;
identifying, in the road segment datastore, a sequence of segments between the start location and the end location that results in a lowest total crash rate compared to a second sequence of segments between the start location and the end location; and
causing the mobile device to display a route on a map of the geographic area that comprises the sequence of segments between the start location and the end location.

### Claim 7 (depends on 6)

The computer-implemented method of claim 6, wherein the start location, the end location, and the second sequence of segments are identified from sensor data collected by the mobile device during a trip in a vehicle.

### Claim 8 (depends on 1)

The computer-implemented method of claim 1, further comprising generating an aggregate risk attribute for the geographic area based on one or more risk attributes in the road segment datastore.

### Claim 9 (depends on 1)

The computer-implemented method of claim 1, wherein the segment attributes for each segment of the plurality of segments comprise static segment attributes and dynamic segment attributes, and wherein the dynamic segment attributes comprise traffic risk attributes and a traffic volume attribute.

### Claim 10 (depends on 9)

The computer-implemented method of claim 9, wherein the static segment attributes include one or more of a segment curvature attribute, a number of lanes attribute, a speed limit attribute, a directionality attribute, a traffic median attribute, a ramp attribute, a lane marker attribute, traffic control signage attributes, a bicycle lane attribute, a pedestrian crossing attribute, or a street light attribute.

### Claim 11 (depends on 9)

The computer-implemented method of claim 9, further comprising:
receiving sensor data collected by a telematics device disposed in a vehicle during a traversal of a physical road segment of the plurality of physical road segments for each traversal of a plurality of traversals of the physical road segment; and
updating the traffic volume attribute for a segment in the road segment datastore that corresponds to the physical road segment based on the sensor data received for each traversal of the plurality of traversals.

### Claim 12 (depends on 11)

The computer-implemented method of claim 11, further comprising:
detecting, for each respective traversal of the plurality of traversals, risk-related driving attributes from the sensor data; and
updating the traffic risk attributes for the segment in the road segment datastore based on the risk-related driving attributes derived for each respective traversal of the plurality of traversals.

### Claim 13 (depends on 12)

The computer-implemented method of claim 12, wherein the risk-related driving attributes include one or more of a speeding attribute, a hard braking attribute, a hard acceleration attribute, a hard cornering attribute, or a distracted driving attribute.

### Claim 14 (depends on 12)

The computer-implemented method of claim 12, wherein updating the traffic risk attributes for the segment in the road segment datastore comprises:
determining, from the risk-related driving attributes derived for each respective traversal of the plurality of traversals, a frequency of each risk-related driving attribute.

### Claim 15

A crash-risk system, comprising:
a crash-risk modeling server system, comprising:
a road segment datastore;
one or more processors; and
a memory storing a set of instructions which, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
receiving digital map data for a geographic area;
instantiating, in the road segment datastore, a plurality of segments, wherein each segment of the plurality of segments has a corresponding physical road segment of a plurality of physical road segments in the geographic area, and wherein each segment of the plurality of segments comprises:
segment attributes derived from the digital map data for the corresponding physical road segment; and
a risk attribute;
determining, from the digital map data, pairs of physical road segments that intersect, wherein each pair of physical road segments that intersect has a corresponding pair of segments in the road segment datastore, and each segment of the plurality of segments is included in at least one pair of segments;
creating, in the road segment datastore, a connection between each segment in each pair of segments, wherein the connection between each pair of segments in the road segment datastore comprises connectivity attributes derived from the digital map data; and
updating, in the road segment datastore, the risk attribute for each segment of the plurality of segments that is connected to one or more paired segments based on:
the segment attributes for the segment;
the segment attributes for the one or more paired segments; and
the connectivity attributes between the segment and the one or more paired segments.

### Claim 16 (depends on 15)

The crash-risk system of claim 15, wherein:
the risk attribute for a segment represents a crash rate per traversal of the segment;
the crash-risk system further comprises an application executing on a mobile device; and
the operations further comprise:
receiving, from the application executing on the mobile device, a start location and an end location within the geographic area;
identifying, in the road segment datastore, a sequence of segments between the start location and the end location that results in a lowest total crash rate compared to a second sequence of segments between the start location and the end location; and
transmitting the sequence of segments between the start location and the end location to the application executing on the mobile device via one or more network connections, wherein in response to receiving the sequence of segments, the application presents the sequence of segments on a map of the geographic area via a display of the mobile device.

### Claim 17 (depends on 16)

The crash-risk system of claim 16, wherein the start location, the end location, and the second sequence of segments are identified from sensor data collected by the application while the mobile device is disposed in a vehicle during a trip.

### Claim 18 (depends on 15)

The crash-risk system of claim 15, wherein:
the crash-risk system further comprises a telematics device disposed in a vehicle; and
the operations further comprise:
receiving sensor data collected by the telematics device during each traversal of a plurality of traversals of a physical road segment of the plurality of physical road segments; and
updating a traffic volume attribute for a segment in the road segment datastore that corresponds to the physical road segment based on the sensor data received for each traversal of the plurality of traversals.

### Claim 19 (depends on 18)

The crash-risk system of claim 18, wherein the operations further comprise:
detecting, for each traversal of the plurality of traversals, risk-related driving attributes from the sensor data; and
updating traffic risk attributes for the segment in the road segment datastore based on the risk-related driving attributes derived for each traversal of the plurality of traversals.

### Claim 20

A non-transitory machine-readable storage medium, including instructions that, when executed by one or more processors of a crash-risk system, cause the one or more processors to perform operations comprising:
instantiating, in a road segment datastore, a plurality of segments, wherein each segment of the plurality of segments has a corresponding physical road segment of a plurality of physical road segments in a geographic area, and wherein each segment of the plurality of segments comprises:
segment attributes derived from digital map data for the corresponding physical road segment; and
a risk attribute;
determining, from the digital map data, pairs of physical road segments that intersect, wherein each pair of physical road segments that intersect has a corresponding pair of segments in the road segment datastore, and each segment of the plurality of segments is included in at least one pair of segments;
creating, in the road segment datastore, a connection between each segment in each pair of segments, wherein the connection between each pair of segments in the road segment datastore comprises connectivity attributes derived from the digital map data; and
updating, in the road segment datastore, the risk attribute for each segment of the plurality of segments that is connected to one or more paired segments based on:
the segment attributes for the segment;
the segment attributes for the one or more paired segments; and
the connectivity attributes between the segment and the one or more paired segments.
