arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2510.04041cs.RO
收藏

SITCOM: 为视觉-语言-动作模型(VLAs)扩展推理时计算量

SITCOM: Scaling Inference-Time COMpute for VLAs

Carnegie Mellon University(卡内基梅隆大学)

Ayudh Saxena, Harsh Shah, Sandeep Routray, Rishi Rajesh Shah, Esha Pahwa

AI总结:

针对VLA模型缺乏前瞻机制、长时域任务误差累积的问题,提出SITCOM框架,通过基于Transformer的动力学模型做多步rollout与奖励选优,将任务完成率从48%提升至72%。

摘要:

由于标注数据采集成本高、对未见环境的泛化能力有限,以及长时域规划存在困难,学习鲁棒的机器人控制策略仍是一项重大挑战。尽管视觉-语言-动作(Vision-Language-Action, VLA)模型通过将自然语言指令落地为单步控制命令提供了一种有前景的解决方案,但它们通常缺乏前瞻机制,在动态任务中容易出现误差累积问题。在本项目中,我们提出了Scaling Inference-Time COMpute for VLAs(SITCOM,面向VLA的推理时计算扩展框架),该框架受模型预测控制算法启发,为任意预训练VLA增加基于模型的 rollout 与基于奖励的轨迹选择功能。SITCOM利用一个已学习的动力学模型模拟多步动作rollout,为真实世界执行选择最优候选方案,将单次推理的VLA转化为鲁棒的长时域规划器。我们开发了一种高效的基于Transformer的动力学模型,该模型在大规模BridgeV2数据集上训练,并在SIMPLER环境中微调以弥补实到仿(Real2Sim)差距,同时使用模拟器的奖励为候选rollout打分。通过在SIMPLER环境的多个任务与设置下开展全面评估,我们证明当搭配良好的奖励函数时,SITCOM借助训练好的动力学模型可将任务完成率从48%显著提升至72%。

英文摘要:

Learning robust robotic control policies remains a major challenge due to the high cost of collecting labeled data, limited generalization to unseen environments, and difficulties in planning over long horizons. While Vision-Language-Action (VLA) models offer a promising solution by grounding natural language instructions into single-step control commands, they often lack mechanisms for lookahead and struggle with compounding errors in dynamic tasks. In this project, we introduce Scaling Inference-Time COMpute for VLAs (SITCOM), a framework that augments any pretrained VLA with model-based rollouts and reward-based trajectory selection, inspired by Model Predictive Control algorithm. SITCOM leverages a learned dynamics model to simulate multi-step action rollouts to select the best candidate plan for real-world execution, transforming one-shot VLAs into robust long-horizon planners. We develop an efficient transformer-based dynamics model trained on large-scale BridgeV2 data and fine-tuned on SIMPLER environments to bridge the Real2Sim gap, and score candidate rollouts using rewards from simulator. Through comprehensive evaluation across multiple tasks and settings in the SIMPLER environment, we demonstrate that SITCOM when combined with a good reward function can significantly improve task completion rate from 48% to 72% using trained dynamics model.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑