Libraries required for the pipeline:
Pytorch: pip install torch torchvision Matplotlib: python -m pip install -U matplotlib Scipy:python -m pip install --user numpy scipy matplotlib ipython jupyter pandas sympy nose
Run the abstract_simulator
command: python abstract_simulator.py The following data documents wiil be generated: reward_comps.csv,cost_comps.csv,successrate_comps.csv,peril_costs.csv,peril_rewards.csv,peril_successrates.csv,plearn_rewards.csv, plearn_costs.csv,rlearn_rewards.csv,rlearn_rewards.csv,rlearn_costs.csv The following plots will be generated: cost_comparison.pdf,cost_curve.pdf,cost_curveband.pdf, reward_comparison.pdf,reward_curve.pdf,reward_curveband.pdf,
Module files:
dtio.py : functions for data output and plot reasoner.py: functions of the reasoner interaction.py: functions of the POMDP planner perception.py: functions of the classifier
POMDP Test
To test the POMDP run the folwoing command: python pomdp_tester.py>results.txt
The results.txt will contain the actions and results of each instance. Three plots of "pomdp_r.pdf", "pomdp_c.pdf" and "pomdp_s.pdf" will be generated.