Patent Issued for Personalized driving risk modeling and estimation system and methods (USPTO 11578990): Allstate Insurance Company

2023 MAR 06 (NewsRx) -- By a News Reporter-Staff News Editor at Insurance Daily News -- Allstate Insurance Company (Northbrook, Illinois, United States) has been issued patent number 11578990, according to news reporting originating out of Alexandria, Virginia, by NewsRx editors.

The patent’s inventors are Chintakindi, Sunil (Menlo Park, CA, US), Kodali, Anuradha (Fremont, CA, US).

This patent was filed on October 30, 2020 and was published online on February 14, 2023.

From the background information supplied by the inventors, news correspondents obtained the following quote: “Driver profiling is shifting from traditional driver characteristics to driving attributes for control system modeling. Understanding driver behaviors in risk-context and the corresponding driving sensitivity to the contextual environment is important for risk modeling and crash causation. There is a need for modeling risk predictive behaviors for dangerous driver behaviors and the corresponding risk scene environment. The driving risk modeling and estimation systems and methods can be helpful for controlling vehicle, coaching drivers, and warning drivers that interact with other human drivers on the road in autonomous, semi-autonomous, or traditional vehicle control systems in the vehicles.

“Additionally, these driver behavior-based profiles can be used for insurance purposes.

“Driver behavior-based profiling can be effective for providing and determining accurate personalized insurance premiums, driver coaching, rewarding safe drivers, designing driver-system handoff in semi-autonomous vehicles, and autonomous systems or advanced driver-assistance system (ADAS).”

Supplementing the background information on this patent, NewsRx reporters also obtained the inventors’ summary information for this patent: “Aspects of this disclosure overcome problems and limitations of the prior art by providing systems and methods to identify driver behaviors crucial for risk-prediction and learn driver behaviors, such as for driver coaching and/or insurance premiums or with autonomous or ADAS systems. Additionally, the systems and methods may help identify how driver behaviors and risk reactions play in the context of scene configuration affecting crash causation and identifying a catalog of risk scenes and driving behaviors. Identifying irregular pattern driving, risky acceleration, hard braking, following vehicles, and lane changes, etc. even during regular driving may be important as these attributes indicate risky driving and are proved to correlate with actual crashes. Similarly, a catalog of high-risk contextual scenes may be useful to identify dangerous cliques by involving driving behaviors, thereby personalizing to the driver-level. This dataset profiling or scoring and bookkeeping of scenes can accommodate multiple applications in autonomous, intelligent transportation, traffic safety, and the insurance context.

“For autonomous applications, this modeling may be extended into two areas: design of control systems (e.g. advanced driver-assistance systems-ADAS) to mimic driving; and a context, driver-system handoff problem as seen in level 3 and level 4 autonomous vehicles for control-shift between human and machine. Importantly, this risk modeling and estimation as disclosed herein is crucial when autonomous vehicles are interacting with regular human drivers sharing the same road. Additionally, the knowledge of the environment and the driver’s influence on risk is important while constructing intelligent transport systems with smart city infrastructure especially for fleet management like Robo taxis.

“Insurance’s conventional approach of determining premiums via driver characteristic factors that are expected to impact future costs is changing fast to include driving with contextual information for reflection of current times. Usage-based insurance is one step towards that direction with “pay as you drive” (PAYD) schemes. With the inclusion of tracking driver behaviors, it can be possible for “pay how you drive” (PHYD) mechanisms. These mechanisms may be as personalized as possible with setting the premium, rewarding safe drivers and providing driver coaching information to the driver.

“Various approaches to driving risk modeling and estimation are presented. In accordance with aspects of this disclosure, a risk-predictive modeling system for providing driving risk modeling and estimation may comprise: one or more vehicle operation sensors configured to record vehicle telematics data from a vehicle associated with a driver; one or more frustration sensors configured to record frustration information indicating a current level of the driver of the vehicle; and one or more telematics devices configured to transmit the recorded vehicle telematics data from the vehicle to a processor. The processor and a memory unit storing computer-executable instructions, which when executed by the processor, may cause the processor to: receive the vehicle telematics data from the one or more telematics devices corresponding to the driver; identify and count one or more driving behaviors from the vehicle telematics data; calculate and determine a driving behavior score dataset from the counted one or more driving behaviors from the vehicle telematics data; calculate and determine a risk-exposure score dataset and a risk map from a historical database storing accident information, traffic data, vehicle volume data, vehicle density data, road characteristic data, or weather data; calculate and determine a driver-risk score dataset based on a weighted average of the driving behavior score dataset and the risk-exposure score dataset; receive factors and characteristics for the driver for a driver characteristics dataset; calculate and determine, using weighting and machine learning, a driver score dataset associated with the driving behavior score dataset, the risk-exposure score dataset, the driver-risk score dataset, and the driver characteristics dataset; calculate and determine a contextual information dataset from one or more of weather data, infrastructure data, traffic data, and vehicle characteristics; receive frustration information from the one or more frustration sensors indicating a level of frustration of the driver; calculate and determine a road frustration index dataset from the frustration information and the level of frustration of the driver; receive real-time vehicle telematics data from the one or more telematics devices corresponding to the driver; identify and count one or more real-time driving behaviors from the real-time vehicle telematics data; calculate and determine a current driving behavior score dataset from the counted one or more real-time driving behaviors from the real-time vehicle telematics data; calculate and determine a trip dynamics dataset based on the real-time vehicle telematics data corresponding to a trip and a time and distance remaining for the trip; calculate and determine, using weighting and machine learning, a real-time driver score dataset associated with the current driving behaviors dataset, the driver score dataset, and the trip dynamics dataset; calculate and determine, using weighting and machine learning, a driver-contextual risk score dataset associated with the contextual information dataset, the road frustration index dataset, and the real-time driver score dataset; calculate and determine a contextual response score dataset that includes scene-configuration information and contextual information of a scene; calculate and determine, using weighting and machine learning, a risk-response readiness dataset associated with the driver-risk score dataset, the driver-contextual risk score dataset, and the road frustration index dataset; and calculate and determine, using weighting and machine learning, a conflict index dataset associated with the contextual response score dataset and the risk-response readiness dataset. An overall driving risk index value may then be determined using the risk-predictive modeling system and includes the driver score dataset, the driver-contextual risk score dataset, and the conflict index dataset.

“In accordance with other aspects of this disclosure, a method for providing driving risk modeling and estimation may comprise the following steps: receive, by a processor and a memory unit storing computer-executable instructions connected to one or more telematics devices, scene-configuration information and contextual information of a scene; calculate and determine, by a conflict index system that includes a conflict index engine connected to the processor, a contextual response score dataset that includes the scene-configuration information and the contextual information of the scene; calculate and determine, by the conflict index engine using weighting and machine learning, a risk-response readiness dataset associated with a driver-risk score dataset based on the vehicle telematics data and a risk map of a route, a driver-contextual risk score dataset based on a real-time telematics data, and a road frustration index dataset based on a frustration level of the driver, wherein the risk-response readiness dataset includes an index of how a set of driving behaviors plus scene-configuration effects a driver’s reactions; and calculate and determine, by the conflict index engine using weighting and machine learning, a conflict index dataset associated with the contextual response score dataset and the risk-response readiness dataset. Further, an overall driving risk index value may be determined using a risk-predictive modeling system and includes a driver score dataset, the driver-contextual risk score dataset, and the conflict index dataset. Further, the one or more telematics devices may be configured to transmit recorded vehicle telematics data from a vehicle to the processor. The scene-configuration information and the contextual information may include one or more of the following: a number of autonomous vehicles, a number of human drivers, a set of pedestrians or others in the scene, a risk map, a number of total vehicles, lane dynamics of the scene, and traffic dynamics of the scene.”

The claims supplied by the inventors are:

“1. A risk-predictive modeling system for providing driving risk modeling and estimation, the system comprising: a computing device comprising: a risk-predictive modeling processor and a non-transitory memory unit storing computer-executable instructions, which when executed by the risk-predictive modeling processor, cause the computing device to: receive vehicle telematics data from a vehicle telematics system, the vehicle telematics system comprising one or more sensors sensing the vehicle telematics data from a vehicle associated with a driver, the vehicle telematics data including at least a speed of the vehicle; calculate and determine a driving behavior score dataset from the vehicle telematics data; calculate and determine a risk-exposure score dataset and a risk map from a historical database storing accident information, traffic data, vehicle volume data, vehicle density data, road characteristic data, or weather data; calculate and determine a driver-risk score dataset based on a weighted average of the driving behavior score dataset and the risk-exposure score dataset; calculate and determine, using weighting and machine learning, a driver score dataset associated with the driving behavior score dataset, the risk-exposure score dataset, the driver-risk score dataset, and a driver characteristics dataset; calculate, in near real-time and based on the vehicle telematics data, a level of frustration associated with the driver; calculate and determine a road frustration index dataset from the level of frustration of the driver; calculate and determine, using weighting and machine learning, a real-time driver score dataset associated with a current driving behaviors dataset, and the driver score dataset; calculate and determine, using weighting and machine learning, a driver-contextual risk score dataset associated with a contextual information dataset, the road frustration index dataset, and the real-time driver score dataset; calculate and determine a contextual response score dataset that includes scene-configuration information and contextual information of a scene; calculate and determine, using weighting and machine learning, a risk-response readiness dataset associated with the driver-risk score dataset, the driver-contextual risk score dataset, and the road frustration index dataset; calculate and determine, using weighting and machine learning, a conflict index dataset associated with the contextual response score dataset and the risk-response readiness dataset; calculate and determine an overall driving risk index value using the risk-predictive modeling system that includes the driver score dataset, the driver-contextual risk score dataset, and the conflict index dataset; and cause, based at least in part on the overall driving risk index value, a vehicle control system of the vehicle to perform a driving action.

“2. The system of claim 1, wherein the risk-predictive modeling processor determines driver coaching information corresponding to the driver based on the overall driving risk index value.

“3. The system of claim 1, wherein the risk-predictive modeling processor provides driver-system handoff in one of semi-autonomous vehicles, autonomous systems, or advanced driver-assistance systems.

“4. The system of claim 1, wherein the scene-configuration information and the contextual information includes at least one of: a number of autonomous vehicles, a number of human drivers, a set of pedestrians or others in the scene, or the risk map.

“5. The system of claim 1, wherein the scene-configuration information and the contextual information includes at least one of: a number of vehicles around, lane dynamics of the scene, or traffic dynamics of the scene.

“6. The system of claim 1, wherein the current driving behaviors dataset includes one or more driving behaviors that include at least one of: improper speed, braking, inattentive, signal violation, improper interaction with others, improper passing, improper turning, avoiding maneuvers, or driving experiences.

“7. The system of claim 1, wherein the driver characteristics dataset includes age, gender, education, and place of residence.

“8. A method for providing driving risk modeling and estimation, the method comprising: receive, from a vehicle telematics system to a risk-predictive modeling processor and a memory unit storing computer-executable instructions connected to the vehicle telematics system, vehicle telematics data, the vehicle telematics system comprising one or more sensors sensing the vehicle telematics data from a vehicle associated with a driver, wherein the vehicle telematics data includes at least a speed of the vehicle; calculate and determine, by the risk-predictive modeling processor using weighting and machine learning, a risk-response readiness dataset associated with a driver-risk score dataset based on the vehicle telematics data and a risk map of a route, a driver-contextual risk score dataset based on a real-time telematics data, and a road frustration index dataset based on a frustration level of the driver, wherein the risk-response readiness dataset includes an index of how a set of driving behaviors plus scene-configuration effects a reactionary action of a driver; calculate and determine, by the risk-predictive modeling processor using weighting and machine learning, a conflict index dataset associated with a contextual response score dataset and the risk-response readiness dataset; calculate and determine an overall driving risk index value using the risk-predictive modeling processor that includes a driver score dataset, the driver-contextual risk score dataset, and the conflict index dataset; and cause, based at least in part on the overall driving risk index value, a vehicle control system of the vehicle to perform a driving action.

“9. The method of claim 8, wherein the risk-predictive modeling processor determines driver coaching information corresponding to the driver based on the overall driving risk index value.

“10. The method of claim 8, wherein the risk-predictive modeling processor provides driver-system handoff in one of semi-autonomous vehicles, autonomous systems, or advanced driver-assistance systems.

“11. The method of claim 8, further including: receive, by a driver score system that includes a driver score engine connected to the risk-predictive modeling processor, the vehicle telematics data corresponding to the driver; identify and count, by the driver score engine, one or more driving behaviors from the vehicle telematics data; calculate and determine, by the driver score engine, a driving behavior score dataset from the one or more driving behaviors from the vehicle telematics data; calculate and determine, by the driver score engine, a risk-exposure score dataset and the risk map from a historical database storing accident information, traffic data, vehicle volume data, vehicle density data, road characteristic data, or weather data; calculate and determine, by the driver score engine, a driver-risk score dataset based on a weighted average of the driving behavior score dataset and the risk-exposure score dataset; receive, by the driver score engine, factors and characteristics for the driver for a driver characteristics dataset; and calculate and determine, by the driver score engine using weighting and machine learning, a driver score dataset associated with the driving behavior score dataset, the risk-exposure score dataset, the driver-risk score dataset, and the driver characteristics dataset.

“12. The method of claim 11, wherein the one or more driving behaviors includes at least one of: improper speed, braking, inattentive, signal violation, improper interaction with others, improper passing, improper turning, avoiding maneuvers, or driving experiences.

“13. The method of claim 11, wherein the driver characteristics dataset includes age, gender, education, and place of residence.

“14. The method of claim 8, further including: calculate and determine, by a driver-contextual risk score system that includes a driver-contextual risk score engine connected to the risk-predictive modeling processor, a contextual information dataset from one or more of weather data, infrastructure data, traffic data, and vehicle characteristics; receive, by the driver-contextual risk score engine, frustration information indicating a level of frustration of the driver; calculate and determine, by the driver-contextual risk score engine, a road frustration index dataset from the level of frustration of the driver; receive, by the driver-contextual risk score engine, real-time vehicle telematics data corresponding to the driver; identify and count, by the driver-contextual risk score engine, one or more real-time driving behaviors from the real-time vehicle telematics data; calculate and determine, by the driver-contextual risk score engine, a current driving behavior score dataset from the one or more real-time driving behaviors from the real-time vehicle telematics data; calculate and determine, by the driver-contextual risk score engine, a trip dynamics dataset based on the real-time vehicle telematics data corresponding to a trip and a time and distance remaining for the trip; calculate and determine, by the driver-contextual risk score engine using weighting and machine learning, a real-time driver score dataset associated with the current driving behavior score dataset, the driver score dataset, and the trip dynamics dataset; and calculate and determine, by the driver-contextual risk score engine using weighting and machine learning, a driver-contextual risk score dataset associated with the contextual information dataset, the road frustration index dataset, and the real-time driver score dataset.”

There are additional claims. Please visit full patent to read further.

For the URL and additional information on this patent, see: Chintakindi, Sunil. Personalized driving risk modeling and estimation system and methods. U.S. Patent Number 11578990, filed October 30, 2020, and published online on February 14, 2023. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(11578990)&db=USPAT&type=ids

(Our reports deliver fact-based news of research and discoveries from around the world.)