chris-bee / docker_kalibr Goto Github PK

MaRS (Modular and Robust Multi-Sensor Fusion) is a robot localization framework that was developed as a direct outcome of the research during my Ph.D. studies. The main C++ library, listed below, can be used to process sensor data in CSV files directly. MaRS also has a ROS1 wrapper that implements the C++ API and allows a real-time and closed-loop capable integration of MaRS for your vehicle.

• MaRS C++ Library
• MaRS ROS1 Wrapper
• MaRS ROS2 Wrapper (Coming soon)

MaRS was used together with other projects such as:

• PeET: Pose Estimation Transformer for Single-View, Multi-Object 6D Pose Estimation soon.
• The AMADEE20 Mars analog mission
• The CNS Flight-Stack for UAV autonomy

The AMADEE20 Mars analog mission led to the development of the AMAZE helicopter (Analog MArs Zone Exploration), which was designed with 18 sensors to create a sophisticated multi-sensor UAV dataset. We called this the INSANE dataset (Increased Number of Sensors for developing Advanced and Novel Estimators), which was published open-access in the A+ International Journal of Robotics Research (IJRR).

A repository with additional tools for the generation of the dataset can be found here:

• INSANE Dataset Tools

The development of the AMAZE helicopter, mentioned above, also led to the development of the full autonomy CNS Flight Stack. The main repository of the flight stack can be found here:

• CNS Flight-Stack