The IBRIDO framework

A software framework for research in Reinforcement Learning-Augmented Model Predictive Control. IBRIDO is made by several packages:

Package	Description
AugMPC	RL-Augmented MPC for Non-Gaited Legged and Hybrid Locomotion
AugMPCEnvs	World interfaces (Isaac Sim, MuJoCo, XBot2) and training environments implementations for AugMPC
AugMPCModels	Demo bundles (models and configurations) for AugMPC
MPCHive	Parallel MPC cluster manager, based on a client-server architecture, for use with data-hungry applications
EigenIPC	Provides the shared-memory tensors and other primitives that all components use to exchange states, commands, references, and metadata
MPCViz	Minimal RViz-based tool for visualizing MPC solutions in real-time

The framework can be installed, configured and experimented with using ibrido-containers.

Associated publications

@misc{patrizi2026rlaugmentedmpcnongaitedlegged,
  author={Patrizi, Andrea and Rizzardo, Carlo and Laurenzi, Arturo and Ruscelli, Francesco and Rossini, Luca and Tsagarakis, Nikos G.},
  journal={IEEE Robotics and Automation Letters}, 
  title={RL-Augmented MPC for Non-Gaited Legged and Hybrid Locomotion}, 
  year={2026},
  volume={},
  number={},
  pages={1-8},
  doi={10.1109/LRA.2026.3675839},
}