SEAL: Simultaneous Exploration and Localization in Multi-Robot Systems

Ehsan Latif, Ramviyas Parasuraman

Abstract

The availability of accurate localization is critical for multi-robot exploration strategies; noisy or inconsistent localization causes failure in meeting exploration objectives. We aim to achieve high localization accuracy with contemporary exploration map belief and vice versa without needing global localization information. This paper proposes a novel simultaneous exploration and localization (SEAL) approach, which uses Gaussian Processes (GP)-based information fusion for maximum exploration while performing communication graph optimization for relative localization. Both these cross-dependent objectives were integrated through the Rao-Blackwellization technique. Distributed linearized convex hull optimization is used to select the next-best unexplored region for distributed exploration. SEAL outperformed cutting-edge methods on exploration and localization performance in extensive ROS-Gazebo simulations, illustrating the practicality of the approach in real-world applications.

Introduction

In multi-robot systems (MRS), exploration and localization are inextricably linked. To construct an accurate map, a robot must maintain precise knowledge of its position; conversely, accurate self-localization often depends on a reliable map of the environment. In GPS-denied or feature-poor environments, such as disaster zones or industrial warehouses, this "chicken-and-egg" problem becomes a significant hurdle. Traditional Simultaneous Localization and Mapping (SLAM) techniques are often computationally expensive and rely heavily on complex feature-matching. The SEAL (Simultaneous Exploration and Localization) framework addresses these challenges by providing a distributed, computationally efficient approach that enables robots to explore uncharted territory while maintaining high-precision relative positioning.

Methodology

The SEAL framework is built upon four primary technical pillars designed to optimize both mapping and positioning in parallel:

  1. Distributed Gaussian Process (GP) Modeling: Instead of traditional occupancy grids, SEAL utilizes GPs to model environment occupancy. Robots share these probabilistic "belief maps" with neighbors, allowing the team to fuse information into a global exploration belief without centralized data processing.
  2. Relative Localization via Graph Optimization: The system leverages the Received Signal Strength Indicator (RSSI) from inter-robot communication to construct a pose estimation graph. This allows the robot swarm to determine relative positions accurately without external sensors.
Overview of the SEAL framework, integrating simulated environments, relative localization graphs, and Gaussian Process-based mapping.
  1. Rao-Blackwellization Integration: This mathematical technique serves as the core integration mechanism, updating the posterior beliefs of both exploration (the map) and localization (the poses) simultaneously. This ensures that refinements in localization directly improve map consistency and vice versa.
  2. Linearized Convex Hull Optimization: To determine optimal movement, the system calculates a convex hull of the observed area. By linearizing this hull, robots can efficiently identify the "next-best" unexplored regions, maximizing coverage while minimizing redundant travel distance.
Detailed configuration for relative localization showing the communication graph and state/velocity exchange between robots.

Experimental Evaluation

The researchers evaluated SEAL using high-fidelity simulations in ROS-Gazebo across two complex environments: the AWS Bookstore, a cluttered indoor space, and a standard House World. SEAL was benchmarked against two prominent strategies: Rapidly-exploring Random Tree (RRT) and Deep Reinforcement Learning (DRL) based exploration.

Results and Discussion

Experimental results indicate that SEAL significantly outperforms existing methods across several critical metrics:

  • Localization Accuracy: SEAL achieved 25% higher localization accuracy (measured by Absolute Trajectory Error) compared to RRT and DGORL baselines.
  • Exploration Efficiency: SEAL robots traveled approximately half the total distance of RRT-based robots to achieve a superior exploration percentage (97% vs. 92%).
  • Map Quality: The Structural Similarity Index (SSIM) of maps produced by SEAL was significantly higher (0.88) than that of RRT (0.69), indicating a more faithful representation of the ground truth.
  • Computational Performance: SEAL reduced time complexity per robot by approximately 35%, maintaining lower CPU and RAM usage than traditional SLAM+RRT methods despite the complexity of its probabilistic updates.

Conclusion

SEAL demonstrates a robust and scalable approach to multi-robot coordination in challenging environments. By integrating GP-based mapping with RSSI-driven graph optimization, the framework achieves high-fidelity results with reduced computational overhead. Future research will explore the application of SEAL in dynamic 3D environments and the integration of advanced loop-closure techniques to further refine long-term map consistency.

Summary updated

Abstract

The availability of accurate localization is critical for multi-robot exploration strategies; noisy or inconsistent localization causes failure in meeting exploration objectives. We aim to achieve high localization accuracy with contemporary exploration map belief and vice versa without needing global localization information. This paper proposes a novel simultaneous exploration and localization (SEAL) approach, which uses Gaussian Processes (GP)-based information fusion for maximum exploration while performing communication graph optimization for relative localization. Both these cross-dependent objectives were integrated through the Rao-Blackwellization technique. Distributed linearized convex hull optimization is used to select the next-best unexplored region for distributed exploration. SEAL outperformed cutting-edge methods on exploration and localization performance in extensive ROS-Gazebo simulations, illustrating the practicality of the approach in real-world applications.
BibTeX
@inproceedings{latif2023seal,
  author = {Ehsan Latif and Ramviyas Parasuraman},
  title = {{SEAL: Simultaneous Exploration and Localization in Multi-Robot Systems}},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems},
  year = {2023},
  url = {https://openreview.net/forum?id=KKQ8pbl6NO},
}
Figure