/CoClusterBridge

Tool for bridging a CPU-based control cluster (e.g. of MPC controllers) with parallel simulation environments (e.g. based on Omniverse Isaac Sim). Aimed at facilitating the implementation of the training phase of Learning-Based Receding Horizon Control schemes.

Primary LanguagePythonGNU General Public License v2.0GPL-2.0

License

icon.svg

If you have a beautiful CPU-based controller (developed, for example, with Trajectory Optimization tools like [horizon-casadi](https://github.com/ADVRHumanoids/horizon) and want to implement a *Learning-Based Receding Horizon Control* approach without having to rewrite your whole controller on GPU, then this might be the right tool for you.

CoClusterBridge is a tool for bridging parallel simulations (typically GPU-based), with a cluster of CPU-based (receding-horizon) controllers and properly handling controllers synchronization and triggering.