Autonomous self-learning artificial intelligence intent system for access control
One embodiment provides an access control system including access control sensors to detect actions performed in a vicinity of an access point, a verification sensor to verify access of the access point, and an electronic processor communicatively coupled to the access control sensors and the verification sensor. The electronic processor is configured to in response to an access intent model satisfying an accuracy condition, deploy the access intent model for the access point and receive a dataset indicating an action performed in the vicinity of the access point. The electronic processor is also configured to predict an access intent to access the access point by applying the access intent model to the dataset and enable access through the access point. The electronic processor is further configured to receive verification data indicating whether the access point is accessed, and automatically assign a label to the dataset based on the verification data.
1. An access control system, the system comprising:
one or more access control sensors configured to detect actions performed by one or more persons in a vicinity of an access point;
a verification sensor configured to verify entrance through the access point by the one or more persons; and
an electronic processor communicatively coupled to the one or more access control sensors and the verification sensor, the electronic processor configured to
in response to an access intent model satisfying an accuracy condition, deploy the access intent model for the access point,
receive, from the one or more access control sensors, a dataset indicating an action performed by one or more persons in the vicinity of the access point,
predict an access intent of the one or more persons to access the access point by applying the access intent model to the dataset,
enable access through the access point,
receive, from the verification sensor, verification data indicating whether the one or more persons entered through the access point, and
automatically assign a label to the dataset based on the verification data, the label indicating whether the one or more persons entered through the access point.
2. The system of claim 1 , wherein the electronic processor is further configured to train the access intent model using historical access data associated with the access point.
3. The system of claim 1 , wherein the electronic processor is further configured to continuously receive access data associated with actions of one or more persons in the vicinity of the access point and continuously train the access intent model with the access data.
4. The system of claim 1 , wherein the electronic processor is further configured to update the access intent model based on the labeled dataset.
5. The system of claim 1 , wherein the electronic processor is further configured to test the access intent model based on the labeled dataset.
6. The system of claim 1 , wherein the electronic processor is configured to enable entrance through the access point for the person when the access intent model predicts that the person has an intent to access the access point.
7. The system of claim 1 , wherein the electronic processor is further configured to
record, using the one or more access control sensors, a second action performed by a second person for a predetermined time period;
determine, using the verification sensor, that the action resulted in an access status; and
assign the second action to the access status in the access intent model.
8. The system of claim 1 , wherein the electronic processor is further configured to
record, using the one or more access control sensors, a third action performed by a second person for a predetermined time period;
generate a prediction of access status based on the action;
update an accuracy metric of the access intent model based on the prediction, wherein the accuracy condition is satisfied when the accuracy metric exceeds an acceptable threshold; and
retrain the access intent model using the third action and the corresponding access status.
9. The system of claim 8 , wherein the electronic processor is configured to operate the access point using a failsafe mechanism when the accuracy condition is not satisfied.
10. The system of claim 9 , wherein the electronic processor is configured to
determine that the accuracy metric has fallen below the acceptable threshold; and
in response to the accuracy metric falling below the acceptable threshold stop using the access intent model to control the access point; and
control the access point using the failsafe mechanism.
11. A method of providing access control, the method comprising:
in response to an access intent model satisfying an accuracy condition, deploying the access intent model for an access point;
receiving, from one or more access control sensors configured to detect actions performed by one or more persons in a vicinity of the access point, a dataset indicating an action performed by the one or more persons in the vicinity of the access point,
predicting, with an electronic processor, an access intent of the one or more persons to access the access point by applying the access intent model to the dataset,
enabling, with the electronic processor, access through the access point,
receiving, from a verification sensor configured to verify entrance through the access point by the one or more persons, verification data indicating whether the one or more persons entered through the access point;
automatically assigning, with the electronic processor, a label to the dataset based on the verification data, the label indicating whether the one or more persons entered through the access point.
12. The method of claim 11 , further comprising training the access intent model using historical access data associated with the access point.
13. The method of claim 11 , further comprising continuously receiving access data associated with actions of one or more persons in the vicinity of the access point and continuously training the access intent model with the access data.
14. The method of claim 11 , further comprising:
updating the access intent model based on the labeled dataset.
15. The method of claim 11 , further comprising:
testing the access intent model based on the labeled dataset.
16. The method of claim 11 , wherein enabling access through the access point for the person includes enabling entrance through the access point for the one or more persons when the access intent model predicts that the one or more persons has an intent to access the access point.
17. The method of claim 11 , further comprising:
recording, using the one or more access control sensors, a second action performed by a second person for a predetermined time period;
determining, using the verification sensor, that the action resulted in an access status; and
assigning, using the electronic processor, the second action to the access status in the access intent model.
18. The method of claim 11 , further comprising:
recording, using the one or more access control sensors, a third action performed by a second person for a predetermined time period;
generating a prediction of access status based on the action;
updating an accuracy metric of the access intent model based on the prediction, wherein the accuracy condition is satisfied when the accuracy metric exceeds an acceptable threshold; and
retraining the access intent model using the third action and the corresponding access status.
19. The method of claim 18 , operating the access point using a failsafe mechanism when the accuracy condition is not satisfied.
20. The system of claim 19 , further comprising
determining that the accuracy metric has fallen below the acceptable threshold; and
in response to the accuracy metric falling below the acceptable threshold stopping using the access intent model to control the access point; and
controlling the access point using the failsafe mechanism.