Objective: The goal is to present to the management the real time tracking of the cars and the loader activities; and to eventually aid the communication of their performance for better management of the resources.
Description The loader is loading into these cars coal, underground. It serves three cars -- each taking turns. These cars then transport and dump the coal at a common destination. The cars travel through the inter-spaces between blocks.
I've sensors outfitted to these three cars and the loader measuring certain dynamics that underlie their operating activities. For instance, I've a sensor installed on the loader streaming data with a signature encoding the loading operation it performs when one of the cars sets itself up for it. I also have a few sensors installed on each of the cars streaming data with distinctive signatures (exhibited when they are getting loaded), and the amperes measured on each wheel as it travels indicating speed. I compute the speed with which the car travel as a function of the rpms recorded on each wheel; I can also determine the turn direction by the virtue of speed difference in the wheels.
Now, I'd like to use some machine learning algorithm to train a model to classify the cars activity(going straight; taking right; taking left; going forward; going backwards; waiting in the queue), individually. But how do I get a broad picture of keeping track of these cars real-time movements and the loader operations by analyzing the streaming data in order to communicate to the management of the happenings. I'm looking for a solution(Robot fleet management, of sorts) in the space of Robotics, Supervised learning and Reinforcement learning, alike.