EP-4736150-A1 - COLLISION PREDICTION BASED ON DIFFERENT TYPES OF IMPACT

Abstract

Implementations claimed and described herein provide systems and methods for generating a prediction for a likelihood of a collision based on a respective collision prediction algorithm for a particular type of impact. In one implementation, determining, by a collision prediction algorithm, a prediction score based on a subset of variables associated with the movement of a mobile device. The prediction score associated with a likelihood that the movement is associated with a particular type of impact associated with the first collision prediction algorithm. A prediction that the movement is not association with the first type of impact when the prediction score is below a threshold score is outputted.

Inventors

• ANTONY, Lucas

Assignees

• Allstate Insurance Company

Dates

Publication Date

20260506

Application Date

20240627

Claims (20)

1. 1. A system comprising: one or more processors; a display with a user interface; and a memory unit storing computer-executable instructions, which when executed by the one or more processors, cause the system to: store kinematic variables associated with movement of a mobile device; extract a first subset of the kinematic variables based on a first time window associated with a first type of impact; receive, by a first collision prediction algorithm, the first subset of variables; determine, by the first collision prediction algorithm, a first prediction score based on the first subset of variables, the first prediction score associated with a likelihood that the movement is associated with the first type of impact, wherein the first type of impact is associated with the first collision prediction algorithm; and output a prediction that the movement is not association with the first type of impact when the first prediction score is below a threshold score.
2. 2. The system of claim 1 , wherein the kinematic variables include at least one of global positioning system (GPS) speed variables, GPS altitude variables, and accelerometer magnitude variables.
3. 3. The system of claim 2, wherein the kinematic variables are convolutional neural network (CNN) variables and the first collision prediction algorithm is a collision prediction CNN.
4. 4. The system of claim 3, wherein the system is further caused to: convert the kinematic variables into features that represent GPS speed and altitude properties and accelerometer magnitude properties associated with the movement of the mobile device, wherein the first prediction score is determined based on the features, and wherein the collision prediction CNN is trained to learn which features contribute most to predicting whether the movement is associated with the first type of impact.
5. 5. The system of claim 2, wherein the global positioning system (GPS) speed variables and the GPS altitude variables are associated with a particular GPS sensor window of seconds relative to impact time and the accelerometer magnitude variables are associated with a particular accelerometer sensor window of seconds relative to impact time, wherein the particular GPS sensor window and the particular accelerometer sensor window are associated with the type of impact.
6. 6. The system of claim 1, wherein the system is further caused to: extract a second su bset of the kinematic variables based on a second time window associated with a second type of impact; receive, by a second collision prediction algorithm, the second subset of variables; determine, by the second collision prediction algorithm, a second prediction score associated with a likelihood that the movement is associated with the second type of impact associated with the second collision prediction algorithm based on the second subset of variables; and output a prediction that the movement is association with the second type of impact when the second prediction score is at or above the threshold score.
7. 7. The system of claim 6, wherein the first type of impact is selected from a group of at least: an amusement park ride impact, a skiing-based impact, a boat- or water-based impact, or an action sports type impact, and the second collision prediction algorithm is trained with true automobile collisions.
8. 8. The system of claim 7, wherein the system is further caused to: aggregating predictions from a plurality of collision prediction algorithms; and output a final prediction that the movement is associated with a collision when the second prediction score is at or above the threshold score and when a third prediction score associated with a third collision prediction algorithm trained with false positive data is below an associated threshold score.
9. 9. The system of claim 1, wherein the system is further caused to: train the first collision prediction algorithm with a training dataset comprising a dataset from counterfactual collisions where movement was not associated with the type of impact or actual historical instances associated with the type of impact.
10. 10. The system of claim 1, wherein the first prediction score is based on a continuous scale between two numbers, wherein the continuous scale correlates to a continuous likelihood of collision.
11. 11. A computer-implemented method comprising: training a first collision prediction algorithm with a training dataset comprising a dataset from counterfactual collisions where movement was not associated with a first type of impact and actual historical instances associated with the first type of impact; storing kinematic variables associated with the movement of a mobile device; extracting a first subset of the kinematic variables based on a first time window associated with the first type of impact; receiving, by the first collision prediction algorithm, the first subset of variables; determining, by the first collision prediction algorithm, a first prediction score based on the first subset of variables, the first prediction score associated with a likelihood that the movement is associated with the first type of impact, wherein the first type of impact is associated with the first collision prediction algorithm; and outputting a prediction that the movement is not association with the first type of impact when the first prediction score is below a threshold score.
12. 12. The computer- implemented method of claim 11 , wherein the kinematic variables include at least one of global positioning system (GPS) speed variables, GPS altitude variables, and accelerometer magnitude variables.
13. 13. The computer-implemented method of claim 12, wherein the kinematic variables are convolutional neural network (CNN) variables and the first collision prediction algorithm is a collision prediction CNN.
14. 14. The computer-implemented method of claim 13, further comprising: converting the kinematic variables into features that represent GPS speed and altitude properties and accelerometer magnitude properties associated with the movement of the mobile device, wherein the first prediction score is determined based on the features, and wherein the collision prediction CNN is trained to leam which features contribute most to predicting whether the movement is associated with the first type of impact.
15. 15. The computer-implemented method of claim 12, wherein the global positioning system (GPS) speed variables and the GPS altitude variables are associated with a particular GPS sensor window of seconds relative to impact time and the accelerometer magnitude variables are associated with a particular accelerometer sensor window of seconds relative to impact time, wherein the particular GPS sensor window and the particular accelerometer sensor window are associated with the type of impact.
16. 16. The computer-implemented method of claim 11, further comprising: extracting a second subset of the kinematic variables based on a second time window associated with a second type of impact; receiving, by a second collision prediction algorithm, the second subset of variables; determining, by the second collision prediction algorithm, a second prediction score associated with a likelihood that the movement is associated with the second type of impact associated with the second collision prediction algorithm based on the second subset of variables; and outputting a prediction that the movement is association with the second type of impact when the second prediction score is at or above the threshold score.
17. 17. The computer- implemented method of claim 16, wherein the first type of impact is selected from a group of at least: an amusement park ride impact, a skiing-based impact, a boat- or water- based impact, or an action sports type impact, and the second collision prediction algorithm is trained with true automobile collisions.
18. 18. The computer-implemented method of claim 17, further comprising: aggregating predictions from a plurality of collision prediction algorithms; and outputting a final prediction that the movement is associated with a collision when the second prediction score is at or above the threshold score and when a third prediction score associated with a third collision prediction algorithm trained with false positive data is below an associated threshold score.
19. 19. A non-transitory computer-readable medium comprising instructions, the instructions, when executed by a computing system, cause the computing system to: store global positioning system (GPS) speed variables, GPS altitude variables, and accelerometer magnitude variables associated with movement of a mobile device; extract a first subset of the global positioning system (GPS) speed variables, the GPS altitude variables, and the accelerometer magnitude variables based on a first time window associated with a first type of impact; receive, by a first collision prediction algorithm, the first subset of variables; determine, by the first collision prediction algorithm, a first prediction score based on the first subset of variables, the first prediction score associated with a likelihood that the movement is associated with the first type of impact, wherein the first type of impact is associated with the first collision prediction algorithm; and output a prediction that the movement is not association with the first type of impact when the first prediction score is below a threshold score.
20. 20. The non-transitory computer-readable medium of claim 19, wherein the global positioning system (GPS) speed variables, the GPS altitude variables, and the accelerometer magnitude variables are convolutional neural network (CNN) variables and the first collision prediction algorithm is a collision prediction CNN.

Description

COLLISION PREDICTION BASED ON DIFFERENT TYPES OF IMPACT FIELD [0001] Aspects of the presently disclosed technology relate generally to predicting collisions and more particularly to collision prediction that distinguishes between different types of impact. BACKGROUND [0002] Collision detection involves determining whether there has been an intersection of two or more objects. There are different ways of determining a set of trigger parameters that would be used for predicting whether or not a vehicular collision has occurred. However, even sophisticated collision detection systems may result in false positives, since there are different types of movement that are similar to collisions when measuring impact. However, such movements may not necessarily be a vehicular collision. With these observations in mind, among others, various aspects of the present disclosure were conceived and developed. SUMMARY [0003] Implementations described and claimed herein address the foregoing by providing systems and methods for generating a prediction for a likelihood of a collision based on a respective collision prediction algorithm for a particular type of impact. In some implementations, kinematic variables associated with movement of a mobile device may be stored. A first subset of the kinematic variables based on a first time window associated with a first type of impact may be extracted. By a first collision prediction algorithm, the first subset of variables may be received. By the first collision prediction algorithm, a first prediction score based on the first subset of variables may be determined. The first prediction score associated with a likelihood that the movement may be associated with the first type of impact. The first type of impact may be associated with the first collision prediction algorithm. A prediction that the movement is not association with the first type of impact when the first prediction score is below a threshold score may be outputted. [0004] In some implementations, a first collision prediction algorithm may be trained by a dataset from counterfactual collisions where movement was not associated with a first type of impact and actual historical instances associated with the first type of impact. Kinematic variables associated with the movement of a mobile device may be stored. A first subset of the kinematic variables may be extracted based on a first time window associated with the first type of impact. By the first collision prediction algorithm, the first subset of variables may be received, and a first prediction score may be determined based on the first subset of variable. The first prediction score may be associated with a likelihood that the movement is associated with the first type of impact. The first type of impact may be associated with the first collision prediction algorithm. A prediction that the movement is not association with the first type of impact when the first prediction score is below a threshold score may be outputted. [0005] In some implementations, global positioning system (GPS) speed variables, GPS altitude variables, and accelerometer magnitude variables associated with movement of a mobile device may be stored. A first subset of the global positioning system (GPS) speed variables, the GPS altitude variables, and the accelerometer magnitude variables based on a first time window associated with a first type of impact may be extracted. By a first collision prediction algorithm, the first subset of variables may be extracted. By the first collision prediction algorithm, a first prediction score based on the first subset of variables may be determined. The first prediction score associated with a likelihood that the movement is associated with the first type of impact. The first type of impact is associated with the first collision prediction algorithm. A prediction that the movement is not association with the first type of impact when the first prediction score is below a threshold score may be outputted. [0006] Other implementations are also described and recited herein. Further, while multiple implementations are disclosed, still other implementations of the presently disclosed technology will become apparent to those skilled in the art from the following detailed description, which shows and describes illustrative implementations of the presently disclosed technology. As will be realized, the presently disclosed technology is capable of modifications in various aspects, all without departing from the spirit and scope of the presently disclosed technology. Accordingly, the drawings and detailed descriptions are to be regarded as illustrative in nature and not limiting. BRIEF DESCRIPTION OF THE DRAWINGS [0007] FIG. 1 illustrates an example diagram showing a prediction system for generating a prediction for a likelihood of a collision based on a respective collision prediction algorithm for a particular type of impact. [0008] FIG. 2 illustrates example gra