###########################################
CS221 Final Project: AI Lander
Authors:
Kongphop Wongpattananukul ([email protected])
Pouya Rezazadeh Kalehbasti ([email protected])
Dong Hee Song ([email protected])
###########################################
This package includes the following files:
- environment.yml
- A conda environment containing all the requirements to run
the contained python files (Python ver. 3.7.4)
- heuristic.py
- Implementation of section 4.1 from the report
- oracle.py
- Implementation of section 4.2 from the report
- linear.py
- Implementation of section 4.3.1 from the report
- linear_based_on_heuristic.py
- Implementation of section 4.3.1 from the report
- deepQlearn.py
- Implementation of section 4.3.2 from the report
- pkgDelivery.py
- Implementation of section 4.3.3 from the report
- takeoff.py
- Implementation of section 4.3.4 from the report
* heuristic.py, oracle.py, linear.py, linear_based_on_heuristic.py, deepQlearn.py
implementations are based on the original lunar lander problem from openAI's gym.
Note that the Q-learning algorithm used in the files is partially based on Assignment
4 (Blackjack) from CS221.
* pkgDelivery.py, takeoff.py are attempts to modify the original environment
to tackle other formats of this problem (e.g. different lander types, different
objectives), and they are derived from the original LunarLander module from
OpenAI's gym package.
The following sample command line argument should run any python function
listed above. The hyperparameters for the included algorithms can be
changed inside each python file.
$ python heuristic.py