US-12623599-B2 - Techniques for utilizing artificial intelligence to identify vulnerable pedestrian behavior

US12623599B2US 12623599 B2US12623599 B2US 12623599B2US-12623599-B2

Abstract

A control system of a vehicle detects a close-call vehicle-pedestrian encounter where the vehicle and a nearby object of concern almost collide and cause and accident, collects data for a previous period before the detected close-call vehicle-pedestrian encounter, the collected data including data captured by a set of perception sensors of the vehicle during the previous period, and transmits the collected data to a computing server configured to train a vehicle-pedestrian encounter model based on the collected data. The computing server then receives vehicle information indicative of a current state of the vehicle and executes the trained vehicle-pedestrian encounter model using the vehicle information to predict a future potential vehicle-pedestrian encounter and transmits encounter information indicative of the future potential vehicle-pedestrian encounter to the control system, which selectively generates an alert indicative of the future potential vehicle-pedestrian encounter for a driver of the vehicle.

Inventors

Paul A Aldighieri
Matthew A Taylor
Shahin Nobari-Tabrizi
Ethan E Bayer

Assignees

FCA US LLC

Dates

Publication Date: 20260512
Application Date: 20240627

Claims (20)

1 . A vehicle-pedestrian encounter prediction system for a vehicle, the vehicle-pedestrian encounter prediction system comprising: a computing server associated with an original equipment manufacturer (OEM) of a plurality of OEM vehicles including the vehicle; and a control system of the vehicle, the control system being configured to: detect a close-call vehicle-pedestrian encounter where the vehicle and a nearby object of concern almost collide and cause and accident; in response to detecting the close-call vehicle-pedestrian encounter: collect data for a previous period before the detected close-call vehicle-pedestrian encounter, the collected data including data captured by a set of perception sensors of the vehicle during the previous period; and transmit, to the computing server, the collected data, wherein the computing server is configured to train a vehicle-pedestrian encounter model based on the collected data; transmit, to the computing server, vehicle information indicative of a current state of the vehicle, wherein the computing server is further configured to execute the trained vehicle-pedestrian encounter model using the vehicle information to predict a future potential vehicle-pedestrian encounter; receive, from the computing server, encounter information indicative of the future potential vehicle-pedestrian encounter; and selectively generate an alert indicative of the future potential vehicle-pedestrian encounter for a driver of the vehicle.
2 . The vehicle-pedestrian encounter prediction system of claim 1 , wherein the set of perception sensors includes a camera system and the collected data includes a video feed captured by the camera system, and wherein the collected data includes the video feed for the previous period.
3 . The vehicle-pedestrian encounter prediction system of claim 2 , wherein the set of perception sensors further includes a location information system and the collected data includes a location of the vehicle and a time and date of the close-call vehicle-pedestrian encounter.
4 . The vehicle-pedestrian encounter prediction system of claim 2 , wherein the computing server is further configured to analyze the collected data to verify or discard the close-call vehicle-pedestrian encounter.
5 . The vehicle-pedestrian encounter prediction system of claim 4 , wherein the computing server is configured to train the vehicle-pedestrian encounter model based the collected data when the close-call vehicle-pedestrian encounter is verified.
6 . The vehicle-pedestrian encounter prediction system of claim 1 , wherein the control system is configured to automatically detect the close-call vehicle-pedestrian encounter when an evasive feature of the vehicle is engaged.
7 . The vehicle-pedestrian encounter prediction system of claim 1 , wherein the control system is configured to detect the close-call vehicle-pedestrian encounter based on input from the driver of the vehicle.
8 . The vehicle-pedestrian encounter prediction system of claim 1 , wherein the vehicle information comprises at least a current location of the vehicle and a current time and date, and wherein the encounter information includes a likelihood and a type of the future potential vehicle-pedestrian encounter.
9 . The vehicle-pedestrian encounter prediction system of claim 8 , wherein the control system is configured to generate the alert when the likelihood exceeds a likelihood threshold and the alert is based on the type of the future potential vehicle-pedestrian encounter.
10 . The vehicle-pedestrian encounter prediction system of claim 9 , wherein the alert comprises at least one of a visual alert, an audio alert, and a haptic alert.
11 . A vehicle-pedestrian encounter prediction method for a vehicle, the vehicle-pedestrian encounter prediction method comprising: detecting, by a control system of the vehicle, a close-call vehicle-pedestrian encounter where the vehicle and a nearby object of concern almost collide and cause and accident; in response to detecting the close-call vehicle-pedestrian encounter: collecting, by the control system, data for a previous period before the detected close-call vehicle-pedestrian encounter, the collected data including data captured by a set of perception sensors of the vehicle during the previous period; and transmitting, by the control system and to a computing server, the collected data, wherein the computing server is associated with an original equipment manufacturer (OEM) of a plurality of OEM vehicles including the vehicle and is configured to train a vehicle-pedestrian encounter model based on the collected data; transmitting, from the control system and to the computing server, vehicle information indicative of a current state of the vehicle, wherein the computing server is further configured to execute the trained vehicle-pedestrian encounter model using the vehicle information to predict a future potential vehicle-pedestrian encounter; receiving, by the control system and from the computing server, encounter information indicative of the future potential vehicle-pedestrian encounter; and selectively generating, by the control system, an alert indicative of the future potential vehicle-pedestrian encounter for a driver of the vehicle.
12 . The vehicle-pedestrian encounter prediction method of claim 11 , wherein the set of perception sensors includes a camera system and the collected data includes a video feed captured by the camera system, and wherein the collected data includes the video feed for the previous period.
13 . The vehicle-pedestrian encounter prediction method of claim 12 , wherein the set of perception sensors further includes a location information system and the collected data includes a location of the vehicle and a time and date of the close-call vehicle-pedestrian encounter.
14 . The vehicle-pedestrian encounter prediction method of claim 12 , wherein the computing server is further configured to analyze the collected data to verify or discard the close-call vehicle-pedestrian encounter.
15 . The vehicle-pedestrian encounter prediction method of claim 14 , wherein the computing server is configured to train the vehicle-pedestrian encounter model based the collected data when the close-call vehicle-pedestrian encounter is verified.
16 . The vehicle-pedestrian encounter prediction method of claim 11 , wherein the detecting of the close-call vehicle-pedestrian encounter comprises automatically detecting, by the control system, the close-call vehicle-pedestrian encounter when an evasive feature of the vehicle is engaged.
17 . The vehicle-pedestrian encounter prediction method of claim 11 , wherein the detecting of the close-call vehicle-pedestrian encounter is based on input from the driver of the vehicle.
18 . The vehicle-pedestrian encounter prediction method of claim 11 , wherein the vehicle information comprises at least a current location of the vehicle and a current time and date, and wherein the encounter information includes a likelihood and a type of the future potential vehicle-pedestrian encounter.
19 . The vehicle-pedestrian encounter prediction method of claim 18 , wherein the generating of the alert is performed when the likelihood exceeds a likelihood threshold and the alert is based on the type of the future potential vehicle-pedestrian encounter.
20 . The vehicle-pedestrian encounter prediction method of claim 19 , wherein the alert comprises at least one of a visual alert, an audio alert, and a haptic alert.

Description

FIELD The present application generally relates to vehicle artificial intelligence (AI) and, more particularly, to techniques for utilizing AI to identify vulnerable pedestrian behavior. BACKGROUND Vehicles and their drivers may experience many “close call” encounters over time where an accident or collision by the vehicle with a pedestrian or another object almost occurs. Each of these “close call” encounters is preceded by an event where nearby objects of concern (pedestrians, animals, other vehicles, etc.) risked entering or did enter the pathway of the vehicle in a careless or otherwise unexpected manner. Conventional solutions to this problem include evasive or responsive vehicle features, such as automated emergency braking (AEB) and forward collision warning (FCW) or avoidance. These evasive features are reactive in that they are engaged immediately before a concern unfolds and thus they do not predict and proactively alert the driver or prime their attention to a potentially upcoming concern. Accordingly, while such conventional encounter avoidance systems for vehicles do work well for their intended purpose, there exists an opportunity for improvement in the relevant art. SUMMARY According to one aspect of the invention, a vehicle-pedestrian encounter prediction system for a vehicle is presented. In one exemplary implementation, the vehicle-pedestrian encounter prediction system comprises a computing server associated with an original equipment manufacturer (OEM) of a plurality of OEM vehicles including the vehicle a control system of the vehicle, the control system being configured to detect a close-call vehicle-pedestrian encounter where the vehicle and a nearby object of concern almost collide and cause and accident, collect data for a previous period before the detected close-call vehicle-pedestrian encounter, the collected data including data captured by a set of perception sensors of the vehicle during the previous period, transmit, to the computing server, the collected data, wherein the computing server is configured to train a vehicle-pedestrian encounter model based on the collected data, transmit, to the computing server, vehicle information indicative of a current state of the vehicle, wherein the computing server is further configured to execute the trained vehicle-pedestrian encounter model using the vehicle information to predict a future potential vehicle-pedestrian encounter, receive, from the computing server, encounter information indicative of the future potential vehicle-pedestrian encounter, and selectively generate an alert indicative of the future potential vehicle-pedestrian encounter for a driver of the vehicle. In some implementations, the set of perception sensors includes a camera system and the collected data includes a video feed captured by the camera system. In some implementations, the set of perception sensors further includes a location information system and the collected data includes a location of the vehicle and a time and date of the close-call vehicle-pedestrian encounter. In some implementations, the computing server is further configured to analyze the collected data to verify or discard the close-call vehicle-pedestrian encounter. In some implementations, the computing server is configured to train the vehicle-pedestrian encounter model based the collected data when the close-call vehicle-pedestrian encounter is verified. In some implementations, the control system is configured to automatically detect the close-call vehicle-pedestrian encounter when an evasive feature of the vehicle is engaged. In some implementations, the control system is configured to detect the close-call vehicle-pedestrian encounter based on input from the driver of the vehicle. In some implementations, the vehicle information comprises at least a current location of the vehicle and a current time and date, and wherein the encounter information includes a likelihood and a type of the future potential vehicle-pedestrian encounter. In some implementations, the control system is configured to generate the alert when the likelihood exceeds a likelihood threshold and the alert is based on the type of the future potential vehicle-pedestrian encounter. In some implementations, the alert comprises at least one of a visual alert, an audio alert, and a haptic alert. According to another aspect of the invention, a vehicle-pedestrian encounter prediction method for a vehicle is presented. In one exemplary implementation, the vehicle-pedestrian encounter prediction method comprises detecting, by a control system of the vehicle, a close-call vehicle-pedestrian encounter where the vehicle and a nearby object of concern almost collide and cause and accident, collecting, by the control system, data for a previous period before the detected close-call vehicle-pedestrian encounter, the collected data including data captured by a set of perception sensors of the vehicle during the previous period, tra