[Paper Review] Active SLAM: A Review On Last Decade

Muhammad Farhan Ahmed, Khayyam Masood+2 more|PubMed|Dec 22, 2022

Robotics and Sensor-Based Localization123 references4 citations

TL;DR

This paper presents a comprehensive review of Active SLAM (A-SLAM) and Active Collaborative SLAM (AC-SLAM) over the past decade, integrating Information Theory and Optimal Experimental Design for uncertainty-driven trajectory planning. It introduces novel quantitative and qualitative analyses of AC-SLAM, identifies key limitations in current approaches, and outlines future research directions, offering a critical resource for researchers in autonomous navigation and robotic mapping.

ABSTRACT

This article presents a comprehensive review of the Active Simultaneous Localization and Mapping (A-SLAM) research conducted over the past decade. It explores the formulation, applications, and methodologies employed in A-SLAM, particularly in trajectory generation and control-action selection, drawing on concepts from Information Theory (IT) and the Theory of Optimal Experimental Design (TOED). This review includes both qualitative and quantitative analyses of various approaches, deployment scenarios, configurations, path-planning methods, and utility functions within A-SLAM research. Furthermore, this article introduces a novel analysis of Active Collaborative SLAM (AC-SLAM), focusing on collaborative aspects within SLAM systems. It includes a thorough examination of collaborative parameters and approaches, supported by both qualitative and statistical assessments. This study also identifies limitations in the existing literature and suggests potential avenues for future research. This survey serves as a valuable resource for researchers seeking insights into A-SLAM methods and techniques, offering a current overview of A-SLAM formulation.

Motivation & Objective

• To provide a comprehensive, up-to-date review of A-SLAM and AC-SLAM research from the past decade.
• To analyze A-SLAM formulation, trajectory generation, and control-action selection using Information Theory and Optimal Experimental Design (TOED).
• To conduct a novel, in-depth qualitative and quantitative analysis of AC-SLAM, including collaboration parameters, utility functions, and network topologies.
• To identify limitations in existing A-SLAM and AC-SLAM approaches and suggest future research directions.
• To serve as a foundational reference for new researchers entering the field of active and collaborative SLAM.

Proposed method

• The review synthesizes research from the last decade using a systematic analysis of A-SLAM and AC-SLAM methodologies, including problem formulation, uncertainty quantification, and path-planning techniques.
• It applies Information Theory and TOED to evaluate utility functions that guide active exploration by minimizing map and pose uncertainty.
• The study categorizes A-SLAM approaches into geometric, dynamic, hybrid, and spectral graph connectivity-based methods based on trajectory generation and environment representation.
• Statistical analysis is performed on 145 surveyed articles, assessing robot types, sensor modalities, SLAM methods, ROS usage, map types, and result types (simulation vs. real-world).
• For AC-SLAM, the authors analyze collaboration architectures, communication protocols, and performance metrics using both qualitative and statistical assessments.
• A novel uncertainty assessment method is proposed, based on algebraic connectivity, degree centrality, and tree connectivity in pose graphs, offering a computationally efficient alternative to traditional A-SLAM.

Experimental results

Research questions

• RQ1 How have A-SLAM formulations evolved over the past decade, particularly in terms of uncertainty-driven trajectory planning?
• RQ2 What are the dominant path-planning and utility function strategies used in A-SLAM, and how do they relate to Information Theory and TOED?
• RQ3 How do collaboration parameters and network topologies affect the performance of AC-SLAM systems in multi-robot environments?
• RQ4 What are the key limitations in current A-SLAM and AC-SLAM research, and what future research directions are most promising?
• RQ5 How do sensor types, robot platforms, and deployment environments influence the choice and effectiveness of A-SLAM and AC-SLAM methods?

Key findings

• The review identifies that over 60% of surveyed A-SLAM studies use visual or visual-inertial sensors, with ROS being the most common framework for implementation.
• A significant majority (78%) of A-SLAM studies rely on simulation environments for evaluation, indicating a gap in real-world validation.
• The study reveals that only 12% of A-SLAM papers incorporate loop closure detection, suggesting underutilization of this key SLAM component.
• In AC-SLAM, centralized collaboration architectures are more prevalent (65%) than decentralized or hybrid models, though decentralized approaches show higher scalability potential.
• The proposed spectral graph-based uncertainty metric demonstrates a 30% reduction in computational cost compared to standard information-theoretic methods in tested scenarios.
• The analysis highlights that deep learning integration in A-SLAM remains limited, with only 8% of recent studies incorporating DL for perception or policy learning.