# SLAM Backends with Objects in Motion: A Unifying Framework and Tutorial

**Authors:** Chih-Yuan Chiu  
**arXiv:** [2207.05043](https://arxiv.org/abs/2207.05043)  
**Date:** 2022-07-11  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2207.05043v7/

## Key points

- Tutorial on SLAM Backends with Objects in Motion
- Unifying framework for extending optimization-based SLAM algorithms to environments with moving objects and features
- Importance of SLAM algorithms in robotics applications such as autonomous navigation and scene mapping
- Introduction of a recently developed optimization-based SLAM backend framework for handling highly dynamic scenes
- Presentation of a novel filtering-based dynamic SLAM algorithm called dynamic EKF SLAM
- Empirical results demonstrating high localization accuracy, accurate estimation of mobile object poses, and precise map generation
- Comprehensive overview of the unifying optimization-based backend framework
- New algorithm effectively handles moving objects and features while maintaining accuracy and efficiency.

## Summary

This paper presents a tutorial on SLAM Backends with Objects in Motion, focusing on a unifying framework that extends existing optimization-based SLAM algorithms to environments with moving objects and features. The authors highlight the importance of SLAM algorithms in supporting various robotics applications, such as autonomous navigation and scene mapping. To address the challenge posed by highly dynamic scenes in these applications, the tutorial introduces a recently developed optimization-based SLAM backend framework that can handle moving objects and features. By incorporating this framework into dynamic SLAM, the authors aim to bridge the gap between recent advancements in the field. They also present a novel filtering-based dynamic SLAM algorithm called dynamic EKF SLAM which is mathematically equivalent to extending the classical EKF SLAM algorithm to dynamic environments. The empirical results obtained using simulated data demonstrate that dynamic EKF SLAM achieves high localization accuracy, accurate estimation of mobile object poses, and precise map generation while exhibiting high efficiency in terms of computational resources. Overall, this tutorial provides valuable insights into addressing the challenges posed by highly dynamic scenes in SLAM applications. It offers a comprehensive overview of the unifying optimization-based backend framework and presents a new algorithm that effectively handles moving objects and features while maintaining accuracy and efficiency.