Pith. sign in

REVIEW 9 cited by

Exploring Spatial Representation to Enhance LLM Reasoning in Aerial Vision-Language Navigation

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.08500 v3 pith:YOU2MYV2 submitted 2024-10-11 cs.RO cs.AI

Yunpeng Gao , Zhigang Wang , Pengfei Han , Linglin Jing , Dong Wang , Bin Zhao This is my paper
classification cs.ROcs.AI
keywords aerialnavigationspatialrepresentationactioncomplexenhanceenvironments
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Aerial Vision-and-Language Navigation (VLN) is a novel task enabling Unmanned Aerial Vehicles (UAVs) to navigate in outdoor environments through natural language instructions and visual cues. However, it remains challenging due to the complex spatial relationships in aerial scenes.In this paper, we propose a training-free, zero-shot framework for aerial VLN tasks, where the large language model (LLM) is leveraged as the agent for action prediction. Specifically, we develop a novel Semantic-Topo-Metric Representation (STMR) to enhance the spatial reasoning capabilities of LLMs. This is achieved by extracting and projecting instruction-related semantic masks onto a top-down map, which presents spatial and topological information about surrounding landmarks and grows during the navigation process. At each step, a local map centered at the UAV is extracted from the growing top-down map, and transformed into a ma trix representation with distance metrics, serving as the text prompt to LLM for action prediction in response to the given instruction. Experiments conducted in real and simulation environments have proved the effectiveness and robustness of our method, achieving absolute success rate improvements of 26.8% and 5.8% over current state-of-the-art methods on simple and complex navigation tasks, respectively. The dataset and code will be released soon.

Discussion (0). Sign in to comment.

Forward citations

Cited by 9 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Parse, Search, and Confirmation: Training-Free Aerial Vision-and-Dialog Navigation with Chain-of-Thought Reasoning and Structured Spatial Memory

    cs.CV 2026-07 conditional novelty 6.5 of 10

    A three-stage parse-search-confirm MLLM pipeline plus structured spatial memory sets training-free SOTA on AVDN, matching or beating several supervised methods on ANDH and ANDH-Full.

  2. Deliberate Before You Fly: Vision-Guided Spatial Deliberation for UAV See-and-Reach Navigation

    cs.RO 2026-08 conditional novelty 6.0 of 10

    DBFly improves UAV see-and-reach success by adding explicit spatial reasoning steps (direction, diagnosis, maneuver, stop) before predicting waypoints, reporting a 25-point gain over the prior SOTA on the UAV-VLN-FOV ...

  3. AeroDuo: Aerial Duo for UAV-based Vision and Language Navigation

    cs.CV 2025-08 conditional novelty 6.0 of 10

    Two drones at different altitudes, one guided by a vision-language model and one by a local navigator, reach targets more often than single-drone baselines on a new UAV navigation benchmark.

  4. UAV-ON: A Benchmark for Open-World Object Goal Navigation with Aerial Agents

    cs.RO 2025-08 unverdicted novelty 6.0 of 10

    UAV-ON is a new benchmark of 14 Unreal Engine environments with 1270 annotated objects that tests whether aerial agents can navigate to goals described by semantic instance-level instructions.

  5. A Comprehensive Survey and Systematic Real-World Evaluation of Embodied Vision-and-Language Navigation

    cs.RO 2026-07 accept novelty 5.5 of 10

    VLN methods show a large sim-to-real gap; a hierarchical system reaches 51% real-world success versus 22% for a monolithic RGB-only system across ten physical scenes.

  6. PINA: Prompt Injection Attack against Navigation Agents

    cs.CR 2026-01 conditional novelty 5.0 of 10

    PINA automatically crafts prompt-injection texts that make LLM navigation agents fail their missions, reporting 75-100% attack success on indoor and outdoor agents.

  7. TopoNav: Topological Graphs as a Key Enabler for Advanced Object Navigation

    cs.RO 2025-09 conditional novelty 5.0 of 10

    A zero-shot object navigation system that builds a text-based topological memory graph, queried by GPT-4o, reports state-of-the-art success rates of 60.1% on HM3D and 45.5% on MP3D.

  8. SkyVLN: Vision-and-Language Navigation and NMPC Control for UAVs in Urban Environments

    cs.RO 2025-07 conditional novelty 5.0 of 10

    An LLM-and-NMPC drone navigation framework that reports 42.4% success on unseen AVDN test data, versus 16.6% for NavGPT, using spatial verbalization and a path memory graph.

  9. Fly0: Persistent Metric Anchoring for Zero-Shot Aerial Vision-Language Navigation

    cs.RO 2026-02 conditional novelty 4.0 of 10

    A zero-shot aerial VLN system that has an MLLM output only 2D image coordinates, then uses depth unprojection and Ego-Planner to navigate, reporting >20 percentage-point SR gains and 31–37% NE reductions over baselines.

Pith tools