Towards Real-time Semantic RGB-D SLAM in Dynamic Environments

TL;DR

This paper introduces a real-time semantic RGB-D SLAM system capable of detecting known and unknown moving objects in dynamic environments, using efficient segmentation and geometry modules to maintain accuracy on low-power platforms.

Contribution

It presents a novel real-time SLAM system that handles dynamic scenes by combining semantic segmentation on keyframes with an efficient geometry-based unknown object detection method.

Findings

Operates in real-time on low-power embedded platforms.

Achieves high localization accuracy in dynamic environments.

Effectively detects both known and unknown moving objects.

Abstract

Most of the existing visual SLAM methods heavily rely on a static world assumption and easily fail in dynamic environments. Some recent works eliminate the influence of dynamic objects by introducing deep learning-based semantic information to SLAM systems. However such methods suffer from high computational cost and cannot handle unknown objects. In this paper, we propose a real-time semantic RGB-D SLAM system for dynamic environments that is capable of detecting both known and unknown moving objects. To reduce the computational cost, we only perform semantic segmentation on keyframes to remove known dynamic objects, and maintain a static map for robust camera tracking. Furthermore, we propose an efficient geometry module to detect unknown moving objects by clustering the depth image into a few regions and identifying the dynamic regions via their reprojection errors. The proposed method is evaluated on public datasets and real-world conditions. To the best of our knowledge, it is one of the first semantic RGB-D SLAM systems that run in real-time on a low-power embedded platform and provide high localization accuracy in dynamic environments.