Driving · 2026-11-13
What World Model Does Active Object Search Need? A Study under Imperfect Detection
Kyeongryeol Go
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
The authors introduce Bayesian Active Miss-aware Object Search (BAMOS), a belief-based world model for active object search under imperfect detection. It predicts detection likelihood (not full scene reconstruction) using a target belief (location+size), an occupancy belief (for visibility), and fitted detector error rates. Ablations in Isaac Sim and HSSD show target belief and its update mechanisms are critical, with contributions growing as detector errors increase. Code is available.
Why it matters
Active object search requires deciding where to look next and when to stop from imperfect (false positive/negative) observations. This work isolates which “world model” components matter for such decisions, and shows that decision-relevant probabilistic beliefs about the target and detection updates can drive performance even when the detector is uncertain. It provides a structured, ablation-based answer to what representations are needed for reliable search under imperfect perception.
Method
- Introduce BAMOS: a Bayesian belief model that predicts the probability a view detects the target using a target belief over (location, size), an occupancy belief for visibility, and fitted detector error rates.
- Compute view-selection utility via information gain I(Z_q;X|D_t) and use a confirming view plus a belief-stop threshold (belief stop) before declaring a cell.
- Run controlled ablations in Isaac Sim and HSSD with two open-vocabulary detectors, replacing components (target belief, occupancy belief, non-detection update, mask position) and injecting detector errors.
Limitation
The paper’s assumptions include a static scene, a known empty layout with unknown furniture, exact depth and camera poses (within sensor range), and a single target object with known category.
Abstract (from arXiv)
In active object search, a robot must decide where to look and when to stop. Imperfect detection complicates both decisions: a non-detection may indicate occlusion, an out-of-view target, or a false negative, while a detection may be a false positive. This raises the question of which representations a world model needs to support reliable search under imperfect detection. We introduce Bayesian Active Miss-aware Object Search (BAMOS) to investigate how target belief, occupancy belief, and detector errors affect search decisions. Through comprehensive ablations in Isaac Sim and HSSD houses, we examine the roles of these representations and their update mechanisms across different detector configurations. Notably, target belief and its update mechanisms are critical to search performance, with their contribution generally increasing as detection errors become more prevalent. These findings highlight the importance of maintaining and updating target beliefs to support reliable active search under uncertain detector observations, providing insights into the design of world models for embodied agents. Code is available in https://github.com/gokyeongryeol/BAMOS.
Related papers
- Acting from Belief, Looking When Needed: A Bayesian Spatial World Model for Navigation under Intermittent Perception
- SearchWorld: Spatial Value-Grounded Imagination for UAV Object Search via World Models
- What Must a World Model Distinguish for Planning?
- ActiveWAM: Evidence-Aware Active Vision for World-Action Models
- PASA-KV: Attacking Action-Facing Visual Routes in World Action Models
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First announced by arXiv: 2026-11-13 · Indexed by Pub-AI: 2026-11-13 UTC
What World Model Does Active Object Search Need? A Study under Imperfect Detection
First announced by arXiv: 2026-11-13 · Original record ↗
- arxiv
- 2611.15689
Explicit source versions (1)
- v1 · Date not supplied
- Preprint available
Source date: 2026-11-13 · First detected: 2026-11-13 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · RSS guid=oai:arXiv.org:2611.15689v1; arxiv:announce_type=new; item pubDate (official metadata)
Abstract · arxiv
In active object search, a robot must decide where to look and when to stop. Imperfect detection complicates both decisions: a non-detection may indicate occlusion, an out-of-view target, or a false negative, while a detection may be a false positive. This raises the question of which representations a world model needs to support reliable search under imperfect detection. We introduce Bayesian Active Miss-aware Object Search (BAMOS) to investigate how target belief, occupancy belief, and detector errors affect search decisions. Through comprehensive ablations in Isaac Sim and HSSD houses, we examine the roles of these representations and their update mechanisms across different detector configurations. Notably, target belief and its update mechanisms are critical to search performance, with their contribution generally increasing as detection errors become more prevalent. These findings highlight the importance of maintaining and updating target beliefs to support reliable active search under uncertain detector observations, providing insights into the design of world models for embodied agents. Code is available in https://github.com/gokyeongryeol/BAMOS.
