This repo contains all code for the final project. Please see the final report for additional details. The video_examples
directory contains sampled trajectories from various versions of the proposed model.
The main files are:
step_dataset.py
- data loader for simulation datasets.step_train.py
- script to train model.step_test.py
- script to evaluate model (step_test_anomaly.py
is modified version to evaluate classification based on likelihood of a trajectory andstep_test_single.py
is modified version to evaluate the MLP recurrent module baseline).
All other directories contain a readme describing the contained files.