Driving · 2026-10-08

SearchWorld: Spatial Value-Grounded Imagination for UAV Object Search via World Models

Yatai Ji, Zhengqiu Zhu, Yong Zhao, Yue Hu, Fanglong Yao, Chen Gao, Pengfei Zhu, Quanjun Yin

AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.

TL;DR

SearchWorld is a recurrent state-space world model for UAV object search that links explicit BEV exploration/obstacle memory with a decoded task-aware spatial value layer. A cognition–action network refines expert-imitation movement policies using value-guided imagination rollouts without training a separate scalar critic. On UAV-ON, the authors report 23.80% success rate (SR), 35.52% oracle success rate (OSR), and 19.92% SR on unseen environments.

Why it matters

The authors address UAV object search under partial observability where many reactive, memory-based agents reason only over past observations. SearchWorld grounds latent imagination in explicit spatial representations (exploration, obstacles, and a spatial value layer), enabling prospective planning in a spatially interpretable way rather than relying on hand-designed map readouts or a scalar critic.

Method

Maintains explicit allocentric BEV spatial-cognition memory with deterministic exploration and obstacle layers computed from depth/pose, plus a task-aware spatial value layer decoded from the RSSM latent state.
Uses a task-oriented RSSM world model (RGB + instruction + BEV memory) to roll out imagined future states in latent space via the prior-only transition for planning.
Replaces actor–critic scalar value with a cognition–action network: decoded spatial value guides movement-action selection by pooling predicted value over imagined action footprints; stop follows the imitated task policy.

Limitation

The authors scope claims to the UAV-ON simulation environment (photorealistic Unreal Engine scenes) and note that real-world factors are not present or modeled: exact 6-DoF pose, ground-truth depth, and fixed camera intrinsics; benchmark discretized actions with idealized execution; frozen pretrained encoders (DINOv2, SigLIP); no sensor-noise, pose perturbation, action execution error, or cross-domain evaluation; and the BEV memory is 2D with a height threshold that abstracts fine vertical structure. They also flag sim-to-real/cross-simulator robustness as unmeasured and leave noise-injection and transfer studies to future work.

Abstract (from arXiv)

Autonomous unmanned aerial vehicle (UAV) object search involves a closed loop of perception, decision-making, and action under partial observability. Urban environments pose several challenges: large search areas and narrow egocentric views limit coverage, dense 3D geometry constrains safe motion, and open-world instructions require identifying a specific target among distractors. Many existing methods mitigate partial observability through explicit maps or memory representations, yet remain largely reactive, reasoning over past observations without explicitly predicting future states. World models enable prospective reasoning through imagined rollouts. However, image-generating world models can incur high inference latency, while spatially grounded planning remains challenging for latent world models. We propose SearchWorld, a recurrent state-space world model that connects explicit spatial memory with value-guided imagination. The model maintains BEV exploration and obstacle memory and decodes a task-aware spatial value layer to guide search. A cognition-action network uses this learned spatial value prior to improve the policy through imagined rollouts, without training a separate scalar critic. Training progresses from world-model learning to expert imitation and imagination-based exploration refinement. On UAV-ON, SearchWorld improves the success rate to 23.8% (19.5% for the strongest published agent) and raises oracle success to 35.5%, while remaining robust on unseen scenes (19.9% success rate). By grounding imagination in explicit spatial representations, SearchWorld enables UAV agents to plan prospectively rather than react.

Publication record

Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.

First public release: 2026-10-07 · Indexed by Pub-AI: 2026-10-08 UTC

preprint