pith. sign in

arxiv: 2606.22670 · v1 · pith:ELA4ATDBnew · submitted 2026-06-21 · 💻 cs.RO

Semantic-Aware Autonomous Exploration for UAVs in Unknown Indoor Environments

Duc-Thien Nguyen , Ngoc Minh Do , Xiem HoangVan , Thanh Nguyen Canh This is my paper

Pith reviewed 2026-06-26 10:06 UTC · model grok-4.3

classification 💻 cs.RO
keywords UAV explorationsemantic mappingprobabilistic roadmapfrontier selectionRGB-D sensorindoor environmentsautonomous navigation
0
0 comments X

The pith

A semantic reward function added to probabilistic roadmaps lets UAVs reach 90-94 percent coverage in less time than geometry-only planners.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper augments an existing Dynamic Exploration Planner with semantic labels from an RGB-D sensor to build a layered probabilistic roadmap. A new reward term steers frontier selection toward regions containing objects and structures instead of treating all unoccupied space equally. Experiments in multiple simulated indoor scenes show the method covers 90 to 94 percent of the environment while cutting both flight time and path length relative to purely geometric baselines. The approach is implemented in ROS Noetic and Gazebo, demonstrating that semantic cues can be fused incrementally without breaking real-time roadmap updates.

Core claim

By layering semantic information onto the incrementally built Probabilistic Roadmap and scoring candidate frontiers with a semantic reward, the UAV selects paths that prioritize high-value regions, yielding exploration coverage between 90 and 94 percent together with shorter total travel distance and lower elapsed time than conventional geometry-based methods.

What carries the argument

The semantic reward function layered on top of the Dynamic Exploration Planner's Probabilistic Roadmap, which scores frontiers according to the presence of meaningful objects detected by the RGB-D sensor.

If this is right

  • Frontier selection becomes biased toward object-rich zones rather than uniform geometric boundaries.
  • The continuously updated roadmap supports repeated replanning without restarting the entire structure.
  • Simultaneous geometric and semantic mapping occurs from a single sensor stream.
  • Exploration time and distance both decrease in indoor test scenes compared with geometry-only planners.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The resulting map may contain more immediately usable object annotations for downstream tasks such as search or manipulation.
  • If sensor noise increases in real flights the same reward could produce incomplete coverage that a pure geometric planner would avoid.
  • Energy savings from shorter paths could extend mission duration on battery-powered platforms.
  • The method could be combined with existing semantic SLAM pipelines to refine labels on the fly.

Load-bearing premise

Semantic labels supplied by the RGB-D sensor are accurate enough that the added reward term improves information gain without leaving coverage gaps or distorting frontier selection.

What would settle it

A controlled run in which the semantic labels contain realistic noise or the reward produces visibly missed rooms while the geometry-only baseline still reaches higher coverage.

Figures

Figures reproduced from arXiv: 2606.22670 by Duc-Thien Nguyen, Ngoc Minh Do, Thanh Nguyen Canh, Xiem HoangVan.

Figure 1
Figure 1. Figure 1: Overview of the proposed semantic-aware exploration framework. Sensor inputs build the 3D map, which feeds the exploration planner. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Incremental roadmap update process. Green nodes and edges form the current roadmap, the thick red path is the previously executed trajectory, the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The three simulated indoor environments used for evaluation, ordered [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative results comparison in Environment 3. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative results of the proposed method in the three environments. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
read the original abstract

Autonomous exploration in unknown environments requires unmanned aerial vehicles (UAVs) to efficiently generate informative trajectories while simultaneously constructing accurate maps. Although many existing exploration methods rely on geometric information, they often lack semantic awareness, resulting in suboptimal exploration efficiency and limited environmental understanding. To address this limitation, this paper proposes a semantic-aware exploration framework that adds semantic information to a roadmap-based exploration strategy. The proposed method builds on the Dynamic Exploration Planner (DEP), which incrementally constructs a Probabilistic Roadmap (PRM), and augments this roadmap with a semantic layer. A semantic reward function is introduced to prioritize regions containing meaningful objects and structures, enabling the UAV to focus on areas with higher information value. Furthermore, the roadmap is continuously updated to support efficient frontier selection and path planning during exploration. The proposed framework is implemented in ROS Noetic and Gazebo using an RGB-D sensor for simultaneous acquisition of geometric and semantic information. Experimental results in multiple simulated environments demonstrate that the proposed approach achieves exploration coverage rates between 90% and 94% while reducing exploration time and travel distance compared with conventional geometry-based exploration methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes a semantic-aware exploration framework for UAVs that augments the Dynamic Exploration Planner (DEP) by adding a semantic layer to its Probabilistic Roadmap (PRM) construction. A semantic reward function is introduced to prioritize regions containing meaningful objects and structures detected via RGB-D sensors. The approach is implemented in ROS Noetic and Gazebo, with the roadmap continuously updated for frontier selection and path planning. The central empirical claim is that the method achieves 90-94% exploration coverage in simulated indoor environments while reducing time and travel distance relative to conventional geometry-based baselines.

Significance. If the performance claims hold under rigorous validation, the work would offer a practical extension of roadmap-based planners like DEP by incorporating semantic information, potentially improving exploration efficiency and map utility in indoor UAV tasks. The explicit use of standard ROS/Gazebo tooling and incremental PRM updates constitutes a reproducible implementation strength that facilitates follow-on work. However, the current presentation leaves the magnitude of any semantic-driven gains unquantified.

major comments (2)
  1. [Abstract] Abstract: The central claims of 90-94% coverage rates together with reductions in exploration time and travel distance are stated without reference to any data table, figure, error bars, number of trials, or baseline quantitative metrics, rendering the empirical contribution unverifiable from the supplied evidence.
  2. [Method] Method/Results: The semantic reward function is introduced solely by its intended effect on information value and frontier selection; no explicit equation, weighting scheme, or independent derivation is supplied, creating a circularity between the reward definition and the metric it is claimed to improve.
minor comments (1)
  1. [Implementation] The description of RGB-D semantic label acquisition would benefit from explicit mention of the semantic segmentation model or label set used.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive review. We address each major comment below and indicate the revisions we will make to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claims of 90-94% coverage rates together with reductions in exploration time and travel distance are stated without reference to any data table, figure, error bars, number of trials, or baseline quantitative metrics, rendering the empirical contribution unverifiable from the supplied evidence.

    Authors: We agree that the abstract should enable verification of the stated performance claims. In the revised version we will update the abstract to explicitly reference the experimental results section, the relevant figures and tables (including error bars where applicable), the number of independent trials, and the geometry-based baselines used for comparison. revision: yes

  2. Referee: [Method] Method/Results: The semantic reward function is introduced solely by its intended effect on information value and frontier selection; no explicit equation, weighting scheme, or independent derivation is supplied, creating a circularity between the reward definition and the metric it is claimed to improve.

    Authors: We acknowledge that an explicit formulation is required. The revised method section will include the mathematical definition of the semantic reward function, the weighting parameters, and the derivation that links semantic labels obtained from the RGB-D sensor to the information-value term used in frontier selection. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The manuscript presents an empirical engineering contribution: it augments an existing DEP/PRM planner with a semantic reward layer, implements the system in ROS/Gazebo, and reports simulation coverage and timing results. No derivation chain, uniqueness theorem, or first-principles prediction is claimed; the reported 90-94 % coverage figures are direct experimental measurements, not quantities obtained by algebraic reduction or by fitting the same metric the reward is defined to optimize. The semantic reward is an explicit design choice whose effect is measured separately, satisfying the self-contained criterion.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no equations or implementation details, so no free parameters, axioms, or invented entities can be extracted.

pith-pipeline@v0.9.1-grok · 5729 in / 1063 out tokens · 15336 ms · 2026-06-26T10:06:47.773661+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

24 extracted references

  1. [1]

    Safe search and rescue operations based on autonomous robots: A systematic review of the general system architecture,

    A. A. Kareem, A. J. Abid, D. A. Hammood, A. Al-Naji, and J. Chahl, “Safe search and rescue operations based on autonomous robots: A systematic review of the general system architecture,”IEEE Access, 2026

    2026

  2. [2]

    Search and rescue operations in wildfires using unmanned aerial vehicles: A multi-agent deep reinforcement learning approach,

    M. Collignon, A. Perrusqu ´ıa, A. Tsourdos, and W. Guo, “Search and rescue operations in wildfires using unmanned aerial vehicles: A multi-agent deep reinforcement learning approach,”Neurocomputing, p. 131211, 2025

    2025

  3. [3]

    Autonomous exploration in unknown urban environments for unmanned aerial vehicles,

    D. Shim, H. Chung, H. J. Kim, and S. Sastry, “Autonomous exploration in unknown urban environments for unmanned aerial vehicles,” inAIAA Guidance, Navigation, and Control Conference and Exhibit, 2005, p. 6478

    2005

  4. [4]

    Combining autonomous exploration, goal-oriented coordination and task allocation in multi-uav scenarios,

    C. Rasche, C. Stern, W. Richert, L. Kleinjohann, and B. Kleinjohann, “Combining autonomous exploration, goal-oriented coordination and task allocation in multi-uav scenarios,” in2010 Sixth International Conference on Autonomic and Autonomous Systems. IEEE, 2010, pp. 52–57

    2010

  5. [5]

    Efficient autonomous exploration planning of large-scale 3-d environments,

    M. Selin, M. Tiger, D. Duberg, F. Heintz, and P. Jensfelt, “Efficient autonomous exploration planning of large-scale 3-d environments,” IEEE Robotics and Automation Letters, vol. 4, no. 2, pp. 1699–1706, 2019

    2019

  6. [6]

    Fuel: Fast uav exploration using incremental frontier structure and hierarchical planning,

    B. Zhou, Y . Zhang, X. Chen, and S. Shen, “Fuel: Fast uav exploration using incremental frontier structure and hierarchical planning,”IEEE Robotics and Automation Letters, vol. 6, no. 2, pp. 779–786, 2021

    2021

  7. [7]

    Receding horizon

    A. Bircher, M. Kamel, K. Alexis, H. Oleynikova, and R. Siegwart, “Receding horizon” next-best-view” planner for 3d exploration,” in 2016 IEEE international conference on robotics and automation (ICRA). IEEE, 2016, pp. 1462–1468

    2016

  8. [8]

    Tare: A hierarchical frame- work for efficiently exploring complex 3d environments

    C. Cao, H. Zhu, H. Choset, and J. Zhang, “Tare: A hierarchical frame- work for efficiently exploring complex 3d environments.” inRobotics: Science and Systems, vol. 5, 2021, p. 2

    2021

  9. [9]

    T-prm: Temporal probabilistic roadmap for path planning in dynamic environments,

    M. H ¨uppi, L. Bartolomei, R. Mascaro, and M. Chli, “T-prm: Temporal probabilistic roadmap for path planning in dynamic environments,” in 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2022, pp. 10 320–10 327

    2022

  10. [10]

    Fsmp: A frontier-sampling-mixed planner for fast autonomous exploration of complex and large 3-d environments,

    S. Zhang, X. Zhang, Q. Dong, Z. Wang, H. Xi, and J. Yuan, “Fsmp: A frontier-sampling-mixed planner for fast autonomous exploration of complex and large 3-d environments,”IEEE Transactions on Instrumen- tation and Measurement, 2025

    2025

  11. [11]

    Deep reinforce- ment learning-based large-scale robot exploration,

    Y . Cao, R. Zhao, Y . Wang, B. Xiang, and G. Sartoretti, “Deep reinforce- ment learning-based large-scale robot exploration,”IEEE Robotics and Automation Letters, vol. 9, no. 5, pp. 4631–4638, 2024

    2024

  12. [12]

    Octomap: An efficient probabilistic 3d mapping framework based on octrees,

    A. Hornung, K. M. Wurm, M. Bennewitz, C. Stachniss, and W. Burgard, “Octomap: An efficient probabilistic 3d mapping framework based on octrees,”Autonomous Robots, vol. 34, no. 3, pp. 189–206, 2013

    2013

  13. [13]

    Semantics- aware receding horizon planner for object-centric active mapping,

    L. Lu, Y . Zhang, P. Zhou, J. Qi, Y . Pan, C. Fu, and J. Pan, “Semantics- aware receding horizon planner for object-centric active mapping,”IEEE Robotics and Automation Letters, vol. 9, no. 4, pp. 3838–3845, 2024

    2024

  14. [14]

    Semantics-aware exploration and inspection path planning,

    M. Dharmadhikari and K. Alexis, “Semantics-aware exploration and inspection path planning,” in2023 IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 3360–3367

    2023

  15. [15]

    Finding things in the unknown: Semantic object-centric exploration with an mav,

    S. Papatheodorou, N. Funk, D. Tzoumanikas, C. Choi, B. Xu, and S. Leutenegger, “Finding things in the unknown: Semantic object-centric exploration with an mav,” in2023 IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 3339–3345

    2023

  16. [16]

    Semantic octree mapping and Shannon mutual information computation for robot exploration,

    A. Asgharivaskasi and N. Atanasov, “Semantic octree mapping and Shannon mutual information computation for robot exploration,”IEEE Transactions on Robotics, 2023

    2023

  17. [17]

    A frontier-based approach for autonomous exploration,

    B. Yamauchi, “A frontier-based approach for autonomous exploration,” inProceedings 1997 IEEE International Symposium on Computational Intelligence in Robotics and Automation CIRA’97. ’Towards New Com- putational Principles for Robotics and Automation’. IEEE, 1997, pp. 146–151

    1997

  18. [18]

    A survey of path planning of industrial robots based on rapidly exploring random trees,

    S. Luo, M. Zhang, Y . Zhuang, C. Ma, and Q. Li, “A survey of path planning of industrial robots based on rapidly exploring random trees,” Frontiers in Neurorobotics, vol. 17, p. 1268447, 2023

    2023

  19. [19]

    Autonomous uav exploration of dynamic environments via incremental sampling and probabilistic roadmap,

    Z. Xu, D. Deng, and K. Shimada, “Autonomous uav exploration of dynamic environments via incremental sampling and probabilistic roadmap,”IEEE Robotics and Automation Letters, vol. 6, no. 2, pp. 2729–2736, 2021

    2021

  20. [20]

    Sampling-based algorithms for optimal motion planning,

    S. Karaman and E. Frazzoli, “Sampling-based algorithms for optimal motion planning,”The international journal of robotics research, vol. 30, no. 7, pp. 846–894, 2011

    2011

  21. [21]

    Autonomous exploration method for fast unknown environment mapping by using uav equipped with limited fov sensor,

    Y . Zhao, L. Yan, H. Xie, J. Dai, and P. Wei, “Autonomous exploration method for fast unknown environment mapping by using uav equipped with limited fov sensor,”IEEE Transactions on Industrial Electronics, vol. 71, no. 5, pp. 4933–4943, 2023

    2023

  22. [22]

    Real-time volumetric-semantic exploration and map- ping: An uncertainty-aware approach,

    R. P. De Figueiredo, J. le Fevre Sejersen, J. G. Hansen, M. Brand ˜ao, and E. Kayacan, “Real-time volumetric-semantic exploration and map- ping: An uncertainty-aware approach,” in2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2021, pp. 9064–9070

    2021

  23. [23]

    Heuristic-based incre- mental probabilistic roadmap for efficient uav exploration in dynamic environments,

    Z. Xu, C. Suzuki, X. Zhan, and K. Shimada, “Heuristic-based incre- mental probabilistic roadmap for efficient uav exploration in dynamic environments,” in2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 11 832–11 838

    2024

  24. [24]

    S3m: Semantic segmentation sparse mapping for uavs with rgb-d cam- era,

    T. N. Canh, V .-T. Nguyen, X. HoangVan, A. Elibol, and N. Y . Chong, “S3m: Semantic segmentation sparse mapping for uavs with rgb-d cam- era,” in2024 IEEE/SICE International Symposium on System Integration (SII). IEEE, 2024, pp. 899–905

    2024