DSP-SLAM++: A Unified Framework for Multi-Class, High-Fidelity Object SLAM in the Wild

2026-06-24

AI Summary

Existing object-aware SLAM systems struggle to simultaneously achieve real-time performance, support for multiple object categories, and high-fidelity semantically consistent modeling. This work proposes DSP-SLAM++, the first framework to enable low-latency, multi-category, geometrically complete, and semantically consistent object-level SLAM on a general-purpose monocular fisheye–LiDAR platform. By introducing an asynchronous mapping pipeline and a multimodal sensor fusion strategy, the method substantially alleviates bottlenecks in the mapping thread, achieving real-time operation on a 25 Hz multi-category dataset. It reduces the maximum object processing latency by 70% while generating detailed and structurally coherent models across diverse object categories.

Abstract

Existing object-aware SLAM systems force a trade-off between real-time performance, multi-class support, and the generation of high-fidelity, semantically coherent object models. To address this trade-off, we present DSP-SLAM++, which extends the DSP-SLAM framework with an asynchronous mapping pipeline for real-time performance and dedicated sensor fusion adaptations for a monocular fisheye-LiDAR suite. Experiments demonstrate that our system generates fine-grained, geometrically-complete shapes for multiple object classes while eliminating severe mapping thread bottlenecks by reducing maximum object processing latency by up to 70\% compared to the state-of-the-art baseline, enabling robust, real-time performance on a challenging 25 Hz multi-class datasets. This work makes high-fidelity, multi-class object SLAM more practical for real-world applications like autonomous driving and robotic manipulation by enabling its use on platforms with common fisheye-LiDAR sensor setups. The open-source code is available at: [github.com/AUBVRL/DSP-SLAMpp].

Problem

• object SLAM
• real-time performance
• multi-class support
• high-fidelity modeling
• semantic coherence

Innovation

• object SLAM
• multi-class
• high-fidelity reconstruction
• asynchronous mapping
• fisheye-LiDAR fusion