soft-state-tracker is a robust target following framework including state tracking and navigation based on CNNs and MLP that can deal with challenging human following tasks with safety and efficiency.
- models ------ state tracker networks and convolutional auto-encoder network
- tracker ------ state tracker server for evaluate and deploy
- train ------ train scripts to training all networks
- AE_train.py -------
- record.csv ------- record the outputs of two different state-tracker networks in the same test environment for evaluation
- train.ipynb ------- jupyter notebook file to train the convolutional auto-encoder network
- utils.py ------- tools for sorting the samples and rename the every item to corrent sequence
- visualization.py ------ some funtions for visualization