Multi S-Graphs: An Efficient Distributed Semantic-Relational Collaborative SLAM (2401.05152v2)

Abstract: Collaborative Simultaneous Localization and Mapping (CSLAM) is critical to enable multiple robots to operate in complex environments. Most CSLAM techniques rely on raw sensor measurement or low-level features such as keyframe descriptors, which can lead to wrong loop closures due to the lack of deep understanding of the environment. Moreover, the exchange of these measurements and low-level features among the robots requires the transmission of a significant amount of data, which limits the scalability of the system. To overcome these limitations, we present Multi S-Graphs, a decentralized CSLAM system that utilizes high-level semantic-relational information embedded in the four-layered hierarchical and optimizable situational graphs for cooperative map generation and localization in structured environments while minimizing the information exchanged between the robots. To support this, we present a novel room-based descriptor which, along with its connected walls, is used to perform inter-robot loop closures, addressing the challenges of multi-robot kidnapped problem initialization. Multiple experiments in simulated and real environments validate the improvement in accuracy and robustness of the proposed approach while reducing the amount of data exchanged between robots compared to other state-of-the-art approaches. Software available within a docker image: https://github.com/snt-arg/multi_s_graphs_docker

• The paper presents a novel framework integrating semantic data into multi-layered graphs to significantly reduce inter-robot communication.
• It introduces a room-based hybrid descriptor that fuses point cloud details with semantic constructs, improving loop closure detection.
• The system, validated through simulations and real-world tests, achieves lower trajectory errors and robust map reconstructions.

Introduction to Collaborative SLAM

Simultaneous Localization and Mapping, commonly known as SLAM, has paramount importance in the field of robotics, particularly when it comes to navigating uncharted or complex environments. The concept takes a leap forward with Collaborative SLAM (CSLAM), which allows multiple robots to work in unison to generate and share a map of their surroundings while also self-localizing within this shared map. Despite its potential, CSLAM is encumbered by challenges, including error-prone loop closure due to reliance on low-level data and scalability issues stemming from substantial data exchange requirements.

Breakthrough in CSLAM: Multi S-Graphs

In an innovative twist to tackle these hurdles, a noteworthy paper introduces Multi S-Graphs, a decentralized CSLAM system that leverages high-level semantic-relational information to enhance cooperative map-building and localization tasks, while also minimizing communication overhead among robots. This system captures the environment's structural essence through hierarchical situational graphs or S-Graphs, representing environmental features like walls and rooms as entities within a graph. By distilling essential semantic information, the system ensures that the exchange of data is both minimal and meaningful, a strategy that promises scalability as well as efficiency.

Key Contributions of Multi S-Graphs

Multi S-Graphs stand out due to a trio of significant advancements in the world of robotics:

1. A novel framework that artfully integrates semantic information within a multi-layered graph, drastically reducing the amount of data that robots need to exchange for effective collaboration.
2. The introduction of a room-based hybrid descriptor which intelligently combines intricate point cloud details with semantic constructs to improve loop closure detection, thus ensuring precise robotic localization even during the so-called "kidnapped robot" problem where a robot’s initial position is unknown.
3. Validated through rigorous testing in both simulated and real-world environments, this approach showcases a stark improvement in localization accuracy and robustness over other leading CSLAM methods, while diminishing the volume of exchanged data between robots.

Experimental Validation and Impacts

The practical implications of the Multi S-Graphs system were rigorously tested against other cutting-edge CSLAM techniques. In simulations and real-world trials alike, Multi S-Graphs delivered lower absolute trajectory errors and more accurate map reconstructions, all with a fraction of data exchange required by other systems. Notably, these improvements do not come at the cost of versatility, as future enhancements aim to incorporate diverse sensory inputs and broaden the applicability across various robotic platforms and environments.

Concluding Thoughts

The Multi S-Graphs CSLAM system emerges as a milestone in robotics SLAM technology. It balances accuracy and operational efficiency, paving the way for enhanced multi-robot collaboration in complex environments. This system stands to revolutionize the efficiency and scalability of robotic mapping, making it an indispensable asset for future explorations and tasks wherein robots are expected to navigate and understand spaces in a cooperative manner.