Model Predictive Path Integral with Reinforcement Learning baseline

Installation

1. Clone the repository

   git clone --recursive git@github.com:Robot-Learning-Course-Project/MPPI-RL.git
   cd MPPI-RL

2. Create an environment, either using conda or virtualenv

   1. Using conda

      conda env create -f environment.yml
      conda activate mppi-rl

   2. Using virtualenv

      python3.10 -m venv .venv
      source .venv/bin/activate

   Install packages

   pip install -e .
   
   cd mppi_rl/dial_mpc
   pip install -e .

Run our method

1. train an RL policy

   mppi_rl/scripts/brax/train.py

2. visualize the RL policy

   copy model path logs/.../policy_stepxxx to play.py

   mppi_rl/scripts/brax/play.py

3. Run our method

   copy model relative path brax_go2/.../value_stepxxx to mppi_rl/dial_mpc/dial_mpc/examples/unitree_go2_trot_hybrid.yaml

   python mppi_rl/dial_mpc/dial_mpc/core/dial_custom_hybrid.py --example unitree_go2_trot_hybrid

Developer Quick Start

• Run scripts/dev-setup.sh to setup the development environment