Snake bot

The code tagged here as v1.0.0 was used to run snake robot called a during the PyCon SK 2018 programming contest and finished at second place.

Features:

• tracks game state such as movements of snakes, scores, frame rate
• searches for the direction of most food using breadth-first-search
• runs an iterative-deepening minimax search for selecting next move

Known bugs:

• tends to trap itself in a small space :)

Possible improvements:

• modify the BFS to find out whether next move will partition the space and try to avoid it if it does - this should make the snake not to trap itself
• use alpha-beta pruning or another pruning strategy to optimize the minimax search
• cache game states when running the iterative search (currently some states are recomputed)

Setup

1. Clone this repository

git clone https://github.com/martin-sucha/pyconsk-2018-snakepit-bot
export BOTPATH="$(pwd -P)/pyconsk-2018-snakepit-bot"

2. Install snakepit-game from https://github.com/pyconsk/snakepit-game

git clone https://github.com/pyconsk/snakepit-game
export GAMEPATH="${pwd -P}/snakepit-game"
cd "$GAMEPATH"
python3.6 -m venv env
source env/bin/activate
pip install -r doc/requirements.txt
export PYTHONPATH=$(pwd -P)
bin/run.py

3. Run $GAMEPATH/bin/run_robot.py --code "$BOTPATH/asnake.py"

Running tests

# setup venv
cd "$BOTPATH"
python3.6 -m venv env
source env/bin/activate
pip install -r requirements.txt

# run tests
py.test

Measuring performance

There are several helper files (bench.py and prof*.py) that were used for measuring performance.

Running timeit:

python -m timeit -v -s 'import bench, asnake, time' 'bench.advance()'
python -m timeit -v -s 'import bench, asnake, time' 'bench.observe()'
python -m timeit -v -s 'import bench, asnake, time' 'bench.search()'
python -m timeit -v -s 'import bench, asnake, time' 'bench.bfs()'

Running cProfile:

python -m cProfile -s time prof.py
python -m cProfile -s time prof2.py
python -m cProfile -s time prof3.py
python -m cProfile -s time prof4.py