Self-learning gate paddles for safe operation
A system and method for self-learning operation of gate paddles is disclosed. Opening and closing of the gate paddles requires timing and other settings to avoid injury and fare evasion. Self-learning allows a machine learning model to adapt to new data dynamically. The new data captured at a fare gate improves the machine learning model, which can be shared the other similar fare gates within a transit system so that learning disseminates.
1. A self-learning gate system that automatically adjusts for transit users, the self-learning system comprising:
a plurality of sensors configured to capture sensor data for the transit users, wherein input data is formed using the sensor data for the transit users;
a processor configured to execute a machine learning model that generates control data based on the input data;
a barrier, wherein the control data includes instructions for a desired movement of the barrier; and
a barrier actuator coupled to the barrier and configured to cause movement of the barrier based on the control data bounded by predetermined operating guidelines for the barrier,
wherein the machine learning model is trained using:
the input data for different type of transit users, and
a comparison between the desired movement of the barrier and an actual movement of the barrier for different transit users.
2. The self-learning gate system that automatically adjusts for transit users of claim 1 , wherein the input data further comprises fare terms for the transit users.
3. The self-learning gate system that automatically adjusts for transit users of claim 1 , wherein the input data further comprises time and location data of the transit users.
4. The self-learning gate system that automatically adjusts for transit users of claim 1 , wherein the input data further comprises location information of the transit users approaching the barrier.
5. The self-learning gate system that automatically adjusts for transit users of claim 1 , wherein different type of transit users generate different control data unique to themselves.
6. The self-learning gate system that automatically adjusts for transit users of claim 1 , wherein the input data further comprises environmental data comprising at least one of the weather data and station traffic.
7. The self-learning gate system that automatically adjusts for transit users of claim 1 , wherein torque applied to the barrier is limited after changes in the machine learning model.
8. A method for implementing a self-learning fare gate system, the method comprising:
capturing first sensor data using one or more sensors for a first transit user;
forming input data comprising the first sensor data;
generating first control data using a machine learning model based on the first input data, wherein the first control data includes instructions for a desired movement of a barrier of a fare gate;
sending the first control data to a barrier actuator to cause a desired movement of the barrier, wherein the barrier actuator is coupled to the barrier;
determining actual movement of the barrier did not match the desired movement;
forming training data based on a comparison between the desired movement of the barrier and the actual movement of the barrier;
training the machine learning model using the training data;
capturing second sensor data using the one or more sensors for a second transit user;
forming second input data comprising the second sensor data; and
generating second control data using the machine learning model based on the first input data, wherein the second control data is different from the first control data.
9. The method for implementing the self-learning gate system of claim 8 , wherein the first input data further comprises fare ticket data for the first transit user.
10. The method for implementing the self-learning gate system of claim 8 , wherein:
the first and second control data defines timing for opening and closing the barrier, and
timing of the barrier movement for the first user is different from the second user.
11. The method for implementing the self-learning gate system of claim 8 , wherein the first input data further comprises account information for the first transit user.
12. The method for implementing the self-learning gate system of claim 8 , wherein determining the actual movement of the barrier includes:
generating barrier data with a sensor coupled to the barrier actuator, wherein the barrier data is indicative of the actual movement of the barrier.
13. The method for implementing the self-learning gate system of claim 8 , wherein the machine learning model further comprises a neural network.
14. The method for implementing the self-learning gate system of claim 8 , where the second control data sent to the barrier actuator is torque limited after changes in the machine learning model.
15. A self-learning gate system that automatically adjusts for transit users, comprising one or more processors and one or more memories with computer code for:
capturing first sensor data using one or more sensors for a first transit user;
forming input data comprising the first sensor data;
generating first control data using a machine learning model based on the first input data, wherein the first control data includes instructions for a desired movement of a barrier of a fare gate;
sending the first control data to a barrier actuator to cause a desired movement of the barrier, wherein the barrier actuator is coupled to the barrier;
determining actual movement of the barrier did not match the desired movement;
forming training data based on a comparison between the desired movement of the barrier and the actual movement of the barrier;
training the machine learning model using the training data;
capturing second sensor data using the one or more sensors for a second transit user;
forming second input data comprising the second sensor data; and
generating second control data using the machine learning model based on the first input data, wherein the second control data is different from the first control data.
16. The self-learning gate system that automatically adjusts for transit users of claim 15 , wherein the first input data further comprises fare ticket data for the first transit user.
17. The self-learning gate system that automatically adjusts for transit users of claim 15 , wherein the first input data further comprises account information for the first transit user.
18. The self-learning gate system that automatically adjusts for transit users of claim 15 , wherein determining the actual movement of the barrier includes:
generating barrier data with a sensor coupled to the barrier actuator, wherein the barrier data is indicative of the actual movement of the barrier.
19. The self-learning gate system that automatically adjusts for transit users of claim 15 , wherein the machine learning model further comprises a neural network.
20. The self-learning gate system that automatically adjusts for transit users of claim 15 , where the second control data sent to the barrier actuator is torque limited after changes in the machine learning model.