Simultaneous Localization and Mapping on a free-flying space robot using EKF-SLAM 🤖

Project done for the Autonomous Systems course @ IST, Portugal.

This project consists of both localizing an Astrobee robot (3 of which are, at the time of writing this, aboard the ISS) as well as mapping its environment using a finite number of landmarks, through an Extended Kalman Filter.

File Structure

Install required dependencies with pip install -r requirements.txt # NOT YET DONE #

- data
    - bags
    - processed
    - bag.py
- main.py
- ekf_slam.py
- ekf_unknown_correpondences.py
- utils.py

Full breakdown:

> Python EKF-SLAM implementation

- main.py
    - Performs EKF-SLAM and outputs plots of real trajectory and landmark location vs. estimation
- ekf_slam.py
    - Implements the EKF-SLAM with known data association
- ekf_unknown_correpondences.py
    - Implements the EKF-SLAM with data association via a Maximum Likelihood estimator

> Environment map definition (map.py)

> Collected robot data

- data
    - bags
        - Raw data obtained via rosbag recordings in the Astrobee simulator.
    - processed
        - Already-processed data into numpy array format.
- bag.py
    - reads rosbags collected in the robot

> Operate the EKF localization code

- localization
    - localization.py
        - Performs localization on a given map using the EKF
    - ekf.py
        - Implements the Extended Kalman Filter (EKF)
    - registration.py
        - Matches two point-clouds using ICP

> Misc. handy methods (utils.py and plot.py)

- data
    - bags
        Raw data obtained through rosbag recordings in the Astrobee simulator.
    - processed
        Already-processed data into Numpy array format.
    - bag.py
- main.py
    Performs EKF-SLAM and outputs plots for analysis/comparison.
- ekf_slam.py
- ekf_unknown_correpondences.py
    Both implement the SLAM algorithm, with the latter dealing with data association
    via a Maximum Likelihood estimator.
- utils.py