Graph-Based Topological Exploration Planning in Large-Scale 3D Environments

Abstract

Currently, state-of-the-art exploration methods maintain high-resolution map representations in order to optimize exploration goals in each step that maximizes information gain. However, during exploring, those "optimal" selections could quickly become obsolete due to the influx of new information, especially in large-scale environments, and result in high-frequency re-planning that hinders the overall exploration efficiency. In this paper, we propose a graph-based topological planning framework, building a sparse topological map in three-dimensional (3D) space to guide exploration steps with high-level intents so as to render consistent exploration maneuvers. Specifically, this work presents a novel method to estimate 3D space's geometry with convex polyhedrons. Then, the geometry information is utilized to group space into distinctive regions. And those regions are added as nodes into the topological map, directing the exploration process. We compared our method with the state-of-the-art in simulated environments. The proposed method achieves higher space coverage and outperforms exploration efficiency by more than 40% during experiments. Finally, a field experiment was conducted to further evaluate the applicability of our method to empower efficient and robust exploration in real-world environments.

Autonomous Teamed Exploration of Subterranean Environments using Legged and Aerial Robots

“…A rich body of work has focused on the problems of autonomous single and multi-robot exploration [21][22][23][24][25][26][27][28][29][30][31][32][33][34][35]. Early work in single-robot exploration included the sampling of "next-best-views" [25], and the detection of frontiers [26], while recent efforts have focused on powerful planning techniques such as random trees and graphs possibly combined with volumetric calculations [22,23,[36][37][38][39], receding horizon techniques [23], multi-objective optimization [36,37], information-theoretic schemes [40], learning-based methods [41], and approaches that account for the likelihood of accumulating localization drift [22,42]. In multi-robot exploration, the seminal work in [30] presented a strategy for multi-robot coordination exploiting a grid map and a planning policy that tries to minimize the collective exploration time by considering both the cost of reaching a certain frontier cell and the "exploration utility" of each such cell as a function of the number of robots moving to that cell.…”

Exploration with Global Consistency Using Real-Time Re-integration and Active Loop Closure

“…Approaches combining the advantages of frontier-based and sampling-based approaches are presented in [1,7,18]- [22]. [18,19] generate global paths based on frontiers and sample paths locally.…”

“…A two-level framework [21] computes exploration paths coarsely at the global scale and finely around the robot. In [22], sparse topological graphs are built to provide high-level guidance. [1] proposed a hierarchical planning framework based on the incrementally extracted frontiers, which shows high-performance exploration in complex environments.…”

Meeting-Merging-Mission: A Multi-robot Coordinate Framework for Large-Scale Communication-Limited Exploration

“…When all the unobserved aims are eliminated by free space, exploration completes. Yang uses convex polyhedrons to estimate 3D free space in [6]. However, the constrain of convex limits its representation capacity, reducing the eliminating efficiency of unobserved aims.…”