A Software Ecosystem for Research in Reinforcement Learning-based Receding Horizon Control.
Main framework packages:
- LRHControl: Learning-based Receding Horizon Control exploiting Reinforcement Learning.
- CoClusterBridge: tool for bridging a CPU-based control cluster with parallel simulation environments.
- OmniRoboGym: wrapper on top of Omniverse Isaac Sim tailored to legged robotics control with learning techniques.
- SharsorIPCpp: shared matrices built on top of POSIX-IPC and Eigen, shipped with python bindings and NumPy support.
- RHCViz: Ros-powered minimal tool to visualize Receding Horizon Control solutions on RViz in realtime.
Other relevant dependencies:
- IsaacSim: GPU-accelerated robotics simulator for AI.
- horizon: a trajectory optimization tool, tailored to robotics and based on Casadi.
Container deployment utils: ibrido-docker.