Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wang, Dingrui, Liang, Zhihao, Ye, Hongyuan, Sun, Zhexiao, Lu, Zhaowei, Zhang, Yuchen, Zhao, Yuyu, Gao, Yuan, Seegert, Marvin, Schäfer, Finn, Qin, Haotong, Li, Wei, Palmieri, Luigi, Jahncke, Felix, Piccinini, Mattia, Betz, Johannes
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2511.17792
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866914474855759872
author Wang, Dingrui
Liang, Zhihao
Ye, Hongyuan
Sun, Zhexiao
Lu, Zhaowei
Zhang, Yuchen
Zhao, Yuyu
Gao, Yuan
Seegert, Marvin
Schäfer, Finn
Qin, Haotong
Li, Wei
Palmieri, Luigi
Jahncke, Felix
Piccinini, Mattia
Betz, Johannes
author_facet Wang, Dingrui
Liang, Zhihao
Ye, Hongyuan
Sun, Zhexiao
Lu, Zhaowei
Zhang, Yuchen
Zhao, Yuyu
Gao, Yuan
Seegert, Marvin
Schäfer, Finn
Qin, Haotong
Li, Wei
Palmieri, Luigi
Jahncke, Felix
Piccinini, Mattia
Betz, Johannes
contents While recent video world models can generate highly realistic videos, their ability to perform semantic reasoning and planning remains unclear and unquantified. We introduce Target-Bench, the first benchmark that enables comprehensive evaluation of video world models' semantic reasoning, spatial estimation, and planning capabilities. Target-Bench provides 450 robot-collected scenarios spanning 47 semantic categories, with SLAM-based trajectories serving as motion tendency references. Our benchmark reconstructs motion from generated videos with a metric scale recovery mechanism, enabling the evaluation of planning performance with five complementary metrics that focus on target-approaching capability and directional consistency. Our evaluation result shows that the best off-the-shelf model achieves only a 0.341 overall score, revealing a significant gap between realistic visual generation and semantic reasoning in current video world models. Furthermore, we demonstrate that fine-tuning process on a relatively small real-world robot dataset can significantly improve task-level planning performance.
format Preprint
id arxiv_https___arxiv_org_abs_2511_17792
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Target-Bench: Can Video World Models Achieve Mapless Path Planning with Semantic Targets?
Wang, Dingrui
Liang, Zhihao
Ye, Hongyuan
Sun, Zhexiao
Lu, Zhaowei
Zhang, Yuchen
Zhao, Yuyu
Gao, Yuan
Seegert, Marvin
Schäfer, Finn
Qin, Haotong
Li, Wei
Palmieri, Luigi
Jahncke, Felix
Piccinini, Mattia
Betz, Johannes
Computer Vision and Pattern Recognition
Robotics
While recent video world models can generate highly realistic videos, their ability to perform semantic reasoning and planning remains unclear and unquantified. We introduce Target-Bench, the first benchmark that enables comprehensive evaluation of video world models' semantic reasoning, spatial estimation, and planning capabilities. Target-Bench provides 450 robot-collected scenarios spanning 47 semantic categories, with SLAM-based trajectories serving as motion tendency references. Our benchmark reconstructs motion from generated videos with a metric scale recovery mechanism, enabling the evaluation of planning performance with five complementary metrics that focus on target-approaching capability and directional consistency. Our evaluation result shows that the best off-the-shelf model achieves only a 0.341 overall score, revealing a significant gap between realistic visual generation and semantic reasoning in current video world models. Furthermore, we demonstrate that fine-tuning process on a relatively small real-world robot dataset can significantly improve task-level planning performance.
title Target-Bench: Can Video World Models Achieve Mapless Path Planning with Semantic Targets?
topic Computer Vision and Pattern Recognition
Robotics
url https://arxiv.org/abs/2511.17792