This repository provides a pybullet simulation of the TriPed robot. It uses a simplified model with kinematic interfaces to quickly set up walking experiments.
For a more thorough simulation that is capable of computing internal forces, use the Matlab simulation of the TriPed which can be found here
The simulation is also packaged into an OpenAI gym environment for reinforcement learning.
The package is made up of a pybullet TriPed simulation and a gym interface. Both can be installed by calling
git clone https://github.com/TriPed-Robot/triped_gym
cd triped_gym
pip install src/
in the desired installation directory
The current base environment does not offer a reward function or environment (in the sense of ground, obstacles, etc...).
These currently have to be added by defining subclasses.
See test_env.py for reference.
Future additions with agents and full environments are planned as soon as they are finished (published).