Welcome to the landing page for projects supporting our neural control research. These projects focus on data-driven nonlinear optimal control of the physical cartpole robot and our INIvincible F1Tenth race car.
- Hardware Neural Control of CartPole and F1TENTH Race Car
- RPGD: A Small-Batch Parallel Gradient Descent Optimizer with Explorative Resampling for Nonlinear Model Predictive Control
All dependencies between the repositories are automatically managed in the form of Git submodules. For example, if you clone the physical-cartpole repository, it will include the CartPoleSimulation, and CartPoleSimulation will include SI_Toolkit and Control_Toolkit. Unlike other repositories, SI_Toolkit is a Python library and needs to be installed. This happens automatically when running pip install -r requirements.txt.
The dependencies are visualised on the chart below. f1tenth_development_gym and f1tenth_system are accessible through F1Tenth-INI.
