$τ_0$-VLA: a Hierarchical Robot Foundation Model with World-Model-Guided Test-Time Computation
2026-08-17 • Robotics
Robotics
AI summaryⓘ
The authors created a robot control system called τ₀-VLA that breaks down long tasks into smaller parts. Instead of deciding everything at once, their model can take extra time to think more carefully about tricky steps before choosing what to do next. This approach helps the robot make better decisions and improves its ability to complete complex tasks. They trained their model on a large amount of real-world data and showed that spending more time on tough decisions leads to better performance.
robot manipulationhierarchical modelvision-language-action (VLA)high-level policylow-level policytest-time computationworld modelsubtask generationmultimodal co-trainingdistribution shift
Authors
Xiaowei Cai, Yunuo Cai, Bingao Chen, Jingxiao Chen, Zhi Chen, Siyuan Feng, Tengyu Hou, Jingshun Huang, Han Jiang, Runkun Ju, Dong Li, Mingxiang Li, Shaowei Li, Xinchen Li, Yifan Li, Yi Liu, Zhongyuan Liu, Jianlan Luo, Junwen Miao, Ruiqi Ni, Buqing Nie, Mingjie Pan, Xinlin Ren, Jianheng Song, Jiaxu Wang, Peiqi Wang, Sen Wang, Xiaoyan Wang, Dafeng Wei, Dongming Wu, Pengwei Xie, Pu Yang, Hangjian Ye, Xiangyu Yue, Jinyu Zhang, Qinglin Zhang, Xueyong Zhao, Pengfei Zhou, Yue Zhou
Abstract
Long-horizon robot manipulation requires a robot to both execute individual skills reliably and sequence them coherently over extended tasks. Most hierarchical vision-language-action (VLA) models make each such decision with a single forward pass, leaving no mechanism to allocate additional computation to difficult or consequential choices. We introduce $τ_0$-VLA, a hierarchical robot foundation model that formulates high-level subtask generation as a compute-scalable inference problem through world-model-guided test-time computation. At each inference step, the high-level policy uses execution memory to generate a subtask and, when needed, searches over alternatives before committing to its output. A low-level policy then executes the generated subtask across multiple robot embodiments. The policy is trained on 40,115 hours of heterogeneous real-world data with multimodal co-training. Across in-domain and distribution-shifted settings, allocating additional test-time computation substantially improves next-subtask prediction accuracy, and these gains translate into higher closed-loop success on long-horizon robot manipulation tasks.