Wildfire defender

Patent metadata

Field	Value
Publication number	US-11288953-B2
Application number	US-202117192120-A
Country	US
Kind code	B2
Filing date	Mar 4, 2021
Priority date	Sep 27, 2019
Publication date	Mar 29, 2022
Grant date	Mar 29, 2022

Abstract

A method includes accessing a first dataset including aerial imagery data, accessing a second dataset including property boundary data, and identifying property boundaries associated with a geographic area. A plurality of artificial-intelligence (AI) models are applied to the datasets to identify and compute information of interest. Based on the first dataset and constrained by the property boundaries, a building detection model can be applied to identify a building footprint, and a tree detection model can be applied to identify one or more trees. An estimated distance can be determined between each of the trees and a nearest portion of the building footprint as separation data, which can be compared to a defensible space guideline to determine a defensible space adherence score. A wildfire risk map can be generated, including the defensible space adherence score associated with the geographic area.

First claim

What is claimed is:

1. A method comprising:
accessing a first dataset comprising aerial imagery data associated with a geographic area;
accessing a second dataset comprising property boundary data associated with the geographic area;
identifying a plurality of property boundaries associated with the geographic area;
applying a building detection model to identify a building footprint based on the first dataset and constrained by the property boundaries;
applying a tree detection model to identify one or more trees based on the first dataset and constrained by the property boundaries;
determining an estimated distance between each of the one or more trees and a nearest portion of the building footprint as separation data;
comparing the separation data to a defensible space guideline to determine a defensible space adherence score; and
generating a wildfire risk map comprising the defensible space adherence score associated with the geographic area and constrained by the property boundaries.

2. The method of claim 1 , further comprising:
identifying one or more neighboring tree pairs based on a location of each of the one or more trees;
determining an estimated tree-to-tree distance for the one or more neighboring tree pairs; and
incorporating the estimated tree-to-tree distance into the separation data.

3. The method of claim 1 , further comprising:
identifying one or more neighboring properties that share at least one of the property boundaries;
performing a cross-property separation analysis with respect to the one or more neighboring properties; and
incorporating a result of the cross-property separation analysis into the separation data.

4. The method of claim 3 , wherein the cross-property separation analysis comprises determining a shortest distance between the building footprint and a structure on the one or more neighboring properties.

5. The method of claim 3 , wherein the cross-property separation analysis comprises determining a shortest distance between the building footprint and one or more trees on the one or more neighboring properties.

6. The method of claim 3 , wherein the cross-property separation analysis comprises determining an estimated tree-to-tree distance with respect to the one or more trees on the one or more neighboring properties.

7. The method of claim 3 , further comprising:
accessing a third dataset comprising a plurality of geographic features associated with the geographic area; and
predicting a fire path spread pattern between the one or more neighboring properties based on the geographic features identified in the third dataset.

8. The method of claim 7 , wherein the geographic features comprise one or more of: an elevation, a body of water, and a type of ground covering.

9. The method of claim 1 , further comprising:
constructing a three-dimensional model of the geographic area based on the aerial imagery data; and
performing a three-dimensional analysis based on the three-dimensional model to determine the separation data.

10. The method of claim 9 , wherein the first dataset comprises a plurality of height data on a per-pixel basis.

11. The method of claim 9 , further comprising:
determining a size-based component of a wildfire risk score based on a location, area, and height of vegetation captured in the three-dimensional model;
predicting a reduction in the wildfire risk score based on reducing either or both of the area and height of vegetation; and
outputting a vegetation pruning recommendation with the wildfire risk map to illustrate the predicted reduction in the wildfire risk score by performing a size reduction of the vegetation.

12. The method of claim 1 , wherein the first dataset comprises infrared data, and further comprising instructions that when executed by the processing device result in:
identifying one or more dead spots in the one or more trees based on the infrared data;
determining a fire risk adjustment based on the one or more dead spots; and
incorporating the fire risk adjustment into the wildfire risk map.

13. The method of claim 12 , further comprising:
identifying a ground covering moisture content based on the infrared data; and
incorporating a predicted impact of the ground covering moisture content in the wildfire risk map.

14. The method of claim 1 , further comprising:
monitoring for a fire event proximate to the geographic area;
predicting a fire spread path based on the fire event and the wildfire risk map; and
outputting a notification of the fire event and the fire spread path to a user interface.

15. The method of claim 14 , further comprising:
determining a current weather condition and a forecast weather condition between a location of the fire event and the geographic area;
predicting a rate of fire spreading on the fire spread path based on the current weather condition and the forecast weather condition;
predicting a fire arrival time based on the rate of fire spreading; and
outputting the prediction of the fire arrival time with the notification of the fire event and the fire spread path to the user interface.

16. The method of claim 1 , further comprising:
receiving an update to the first dataset;
comparing the update to the first dataset with a previous version of the first dataset;
identifying one or more changes between the previous version of the first dataset and the update to the first dataset; and
modifying the wildfire risk map based on the one or more changes.

Assignees

• Travelers Indemnity Co

Inventors

• Ton-That Hoa
• Dykstra James
• Han John
• Walker Stefanie M
• Amuso Joseph
• Lee George
• Kelsey Kyle J

Classifications

• G08B31/00 Primary

  Predictive alarm systems characterised by extrapolation or other computation using updated historic data · CPC title

• G06F16/909 Primary

  using geographical or spatial information, e.g. location (spatiotemporally dependent retrieval from the web G06F16/9537) · CPC title

• G06V10/80

  Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level (multimodal speaker identification or verification G10L17/10) · CPC title

• G06F18/25

  Fusion techniques · CPC title

• G06T2207/10036

  Multispectral image; Hyperspectral image · CPC title