Multi Robot EM Exploration

This is the implementation of our autonomous exploration algorithm designed for decentralized multi-robot teams, which takes into account map and localization uncertainties of range-sensing mobile robots. Virtual landmarks are used to quantify the combined impact of process noise and sensor noise on map uncertainty. Additionally, we employ an iterative expectation-maximization inspired algorithm to assess the potential outcomes of both a local robot’s and its neighbors’ next-step actions.

• Here we provide a 2D example.
• The Environment and local motion planner used in this repo is developed from our IROS 2023 paper here.

  ├── Multi-Robot-EM-Exploration
     ├── marinenav_env          # IROS 2023 Environment
     ├── nav                    # Exploration and SLAM code
     ├── scripts/test
     │   └── test_multi_SALM.py # 2D example
     └── ...

Dependencies

• numpy
• scipy
• gtsam
• tqdm
• matplotlib

How to use

python3 scripts/test/test_multi_SALM.py