Accompanying code to the publication "Traversing the Reality Gap via Simulator Tuning". Most of this code was written pre-2020 and as such has not been tested recently and is not maintained. If you are having issues running with it please get in contact.
Jack Collins, Ross Brown, Jürgen Leitner and David Howard
Traversing the Reality Gap via Simulator Tuning
Australian Robotics and Automation Association (ARAA) Australasian Conference on Robotics and Automation (ACRA), 2021
This code was developed with Python 3.5 on Ubuntu 16.04. Python libraries required: numpy, scipy optimize, matplotlib, sklearn and skopt.
V-REP: Version 3.5.0*
PyBullet: Version 2.1.1
*Please add vrep.py, vrepConst.py and remoteApi.so files relevant for your distribution of V-Rep to the Software directory.
To execute run main.py
There are three arguments.
- Physics Engine: A string. The choice is between 'PyBullet', 'Bullet278', 'Bullet283', 'ODE' and 'Newton'.
- Experiment number: This is an intiger between 1-6. Where 1 = 1&2, 2 = 3&4, ..., 6 = 11
- Individual/Shared Params: 0=Shared and 1=Individual
The results from execution main.py will be saved in the Results directory. The results are in the form of .csv files. To understand further read the code in logger.py.
Contact
Any questions or comments contact Jack Collins.