BloomBus-AI is the code behind the AI powering the bus arrival time predictions in BloomBus-Client which will be trained on a Nvidia Jetson TX2 Developer Kit and then used in BloomBus-Server.
The AI will focus on predicting the time a bus will arrive at a bus stop, from the data avaiable with the REST API of BloomBus-Server (connected to the BloomBus Firebase DB).
Training can be initiated by using the npm start
command in the Jetson's terminal while in the project directory.
The following reflects the code as it currently stands and is subject to possible change in the near future:
Input Variables
- distance between bus and target bus stop along route
- bus's speed
- number of intersections along route between bus and target bus stop
- number of other bus stops along route between bus and target bus stop
- time bus's speed and location were recorded
Expected Output
- bus arrival time at target bus stop
Once there is a trained model that can attempt to predict bus arrival times, the model will be uploaded to the server hosting BloomBus-Server where the server can call predictArrivalTime(loop, bus, busStop)
to get the bus arrival time predictions for each bus stop.