eSpinn (Evolving SPIking Neural Networks) is a c++ package that simulates Spiking Neural Networks (SNNs) for control problems.
git clone https://github.com/hnqiu/eSpinn.git
cd eSpinn
mkdir build && cd build
cmake ..
makecmake options:
cmake -DCMAKE_BUILD_TYPE=Debug ..
cmake -DNDEBUG=ON .. # if run in Ndebug mode
cmake -DESPINN_VERBOSE=ON .. # if run in verbose debug mode
cmake -DUSE_IN_MATLAB=ON .. # if package is used in MATLAB
cmake -DPYTHON_EXECUTABLE="$HOME/anaconda3/bin/python" .. # if package is compiled with conda env, i.e., the pybind module will be interpreted by the specified Python instance- Boost serialization
# should have already been available on Ubuntu, otherwise: sudo apt install libboost-all-dev - pybind11
Refer to this post for more details on installing pybind11.
conda install -c conda-forge pybind11
- setup
python tasks/sim_flappy.py -t
flappybird.mp4
To use this work, please cite:
@InProceedings{Qiu2020,
author = {Qiu, Huanneng and Garratt, Matthew and Howard, David and Anavatti, Sreenatha},
booktitle = {Proceedings of the 2020 Genetic and Evolutionary Computation Conference},
title = {Towards Crossing the Reality Gap with Evolved Plastic Neurocontrollers},
year = {2020},
address = {New York, NY, USA},
pages = {130-138},
publisher = {Association for Computing Machinery},
series = {GECCO '20},
doi = {10.1145/3377930.3389843},
isbn = {9781450371285},
keywords = {hebbian plasticity, neuroevolution, spiking neural networks, UAV control, evolutionary robotics},
location = {Canc\'{u}n, Mexico},
numpages = {9},
url = {https://doi.org/10.1145/3377930.3389843},
}