Learning to See and Act: Task-Aware Virtual View Exploration for Robotic Manipulation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bai, Yongjie, Wang, Zhouxia, Liu, Yang, Luo, Kaijun, Wen, Yifan, Dai, Mingtong, Chen, Weixing, Chen, Ziliang, Liu, Lingbo, Li, Guanbin, Lin, Liang
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918394546094080
author Bai, Yongjie
Wang, Zhouxia
Liu, Yang
Luo, Kaijun
Wen, Yifan
Dai, Mingtong
Chen, Weixing
Chen, Ziliang
Liu, Lingbo
Li, Guanbin
Lin, Liang
author_facet Bai, Yongjie
Wang, Zhouxia
Liu, Yang
Luo, Kaijun
Wen, Yifan
Dai, Mingtong
Chen, Weixing
Chen, Ziliang
Liu, Lingbo
Li, Guanbin
Lin, Liang
contents Recent vision-language-action (VLA) models for multi-task robot manipulation often rely on fixed camera setups and shared visual encoders, which limit their performance under occlusions and during cross-task transfer. To address these challenges, we propose Task-aware Virtual View Exploration (TVVE), a framework that learns to select task-relevant virtual camera viewpoints and dynamically re-render observations from a reconstructed scene representation using the selected viewpoints. To enable efficient view selection, we train an exploration policy in a pseudo-environment. In addition, we introduce a Task-aware Mixture-of-Experts (TaskMoE) visual encoder that routes visual features to task-specialized experts, mitigating interference in multi-task learning. To evaluate robustness under distribution shifts, we construct RLBench-OG, an out-of-distribution benchmark with visual perturbations and camera pose variations. Experiments on RLBench and RLBench-OG demonstrate that TVVE achieves higher success rates than strong baselines, while real-robot experiments further confirm its robustness to visual disturbances and unseen instructions. Code and visualizations are available at: https://hcplab-sysu.github.io/TAVP.
format Preprint
id arxiv_https___arxiv_org_abs_2508_05186
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning to See and Act: Task-Aware Virtual View Exploration for Robotic Manipulation
Bai, Yongjie
Wang, Zhouxia
Liu, Yang
Luo, Kaijun
Wen, Yifan
Dai, Mingtong
Chen, Weixing
Chen, Ziliang
Liu, Lingbo
Li, Guanbin
Lin, Liang
Robotics
Computer Vision and Pattern Recognition
Recent vision-language-action (VLA) models for multi-task robot manipulation often rely on fixed camera setups and shared visual encoders, which limit their performance under occlusions and during cross-task transfer. To address these challenges, we propose Task-aware Virtual View Exploration (TVVE), a framework that learns to select task-relevant virtual camera viewpoints and dynamically re-render observations from a reconstructed scene representation using the selected viewpoints. To enable efficient view selection, we train an exploration policy in a pseudo-environment. In addition, we introduce a Task-aware Mixture-of-Experts (TaskMoE) visual encoder that routes visual features to task-specialized experts, mitigating interference in multi-task learning. To evaluate robustness under distribution shifts, we construct RLBench-OG, an out-of-distribution benchmark with visual perturbations and camera pose variations. Experiments on RLBench and RLBench-OG demonstrate that TVVE achieves higher success rates than strong baselines, while real-robot experiments further confirm its robustness to visual disturbances and unseen instructions. Code and visualizations are available at: https://hcplab-sysu.github.io/TAVP.
title Learning to See and Act: Task-Aware Virtual View Exploration for Robotic Manipulation
topic Robotics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.05186