This repository contains the implementation of self-assessment extensions for the Stone Soup framework. These extensions are part of our ongoing research on performance monitoring in tracking systems.

For more technical details, please refer to our publication Self-Assessment and Monitoring Module for Tracking Algorithms in the Stone Soup Framework with the corresponding slides.

This publication has won the Second Place Best Paper Award at 2025 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI).

Our proposal introduces a Self-Assessment (SA) module, referred to as Self-Assessor, into the Stone Soup framework. The module enables tracking algorithms to monitor and evaluate their own performance in real-time, facilitating more reliable decision-making in autonomous systems.

If you find this repository useful in your research, please consider citing our work.

@INPROCEEDINGS{griebel2025aduulmstonesoup,
  author={Griebel, Thomas and Wodtko, Thomas and Buchholz, Michael and Dietmayer, Klaus},
  booktitle={2025 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)}, 
  title={Self-Assessment and Monitoring Module for Tracking Algorithms in the Stone Soup Framework}, 
  year={2025},
  volume={},
  number={},
  pages={1-8},
  doi={10.1109/MFI67357.2025.11259162}
}

The following publications are included in the self-assessment framework:

• Kalman Filter Meets Subjective Logic: A Self-Assessing Kalman Filter Using Subjective Logic
• Self-Assessment for Single-Object Tracking in Clutter Using Subjective Logic
• Online Performance Assessment of Multi-Sensor Kalman Filters Based on Subjective Logic

To start developing with our self-assessment extensions, please use Python 3.12 and clone the appropriate branch:

git clone "https://github.com/uulm-mrm/aduulm-stonesoup.git"
cd Stone-Soup
python -m pip install -e ".[dev,aduulm]"

Make sure to check out our self-assessment extensions branch: selfassessment_extensions

If you want to experiment with the Self-Assessor, tutorials can be found here:

👉 Self-Assessor Tutorials

These tutorials allow you to:

• Disturb and manipulate ground truth trajectories
• Disturb and manipulate measurements
• Obtain self-assessment results to detect disturbances

The currently available self-assessment approaches consist of:

• Linear and nonlinear Kalman filter self-assessor

  • 01_KalmanFilterTutorialWithSelfAssessment.py
  • 02_ExtendedKalmanFilterTutorialWithSelfAssessment.py
  • 03_UnscentedKalmanFilterTutorialWithSelfAssessment.py

• Multi-sensor Kalman filter self-assessor

  • 01_MultiSensorKalmanFilterTutorialWithSelfAssessment.py

• Self-assessor for (multi-sensor) single-object tracking in clutter using nearest neighbour association

  • 05_DataAssociation-ClutterWithSelfAssessment.py
  • 05_MultiSensorDataAssociation-ClutterWithSelfAssessment.py

• Self-assessor for (multi-sensor) single-object tracking in clutter using probabilistic data association

  • 07_PDATutorialWithSelfAssessment.py
  • 07_MultiSensorPDATutorialWithSelfAssessment.py

✅ Try out the tutorials and start experimenting with self-assessment for your own tracking setups.

--- Here begins the original Stone Soup README ---

The following section contains the unmodified README from the original Stone Soup project.

Stone Soup is a software project to provide the target tracking and state estimation community with a framework for the development and testing of tracking and state estimation algorithms.

An article is available that details the background to the project, and contains links to sample data.

Please see the Stone Soup documentation for more information.

Please see the tutorials, examples, and demonstrations, which you can also try out on Binder:

Stone Soup is released under MIT License. Please see License for details.