机器人学语言：用视频生成模型当“预知”老师
机器人学做家务，最难的不是记住“把杯子放桌上”，而是当你说“把红杯子放到蓝盘子旁边”时，它得理解颜色、位置、顺序的组合——这叫组合泛化。现有方法要么让机器人死记硬背指令，要么让它自己从零学预测未来画面，结果既慢又笨。InternVLA-A1.5 的做法是：让机器人一边继续学语言问答（保持语义理解），一边用现成的视频生成模型当“预知老师”——老师把未来几秒的关键信息压缩成几个“潜变量”（类似关键词），机器人只需学会猜这些关键词，不用自己生成整段视频。训练时用 120 万条机器人数据和 300 万条多模态数据，测试时扔掉视频部分，只保留轻量推理。结果在 6 个模拟基准上全面领先，真实场景中组合新指令（比如从未见过的颜色-位置搭配）成功率最高，且能执行长序列任务。它不是你明天就能用的产品，但指明了方向：让机器人“借用”大模型的世界知识，而不是什么都从头学。
📄 原文摘要(英文)
Unified models for robot manipulation aim to equip one policy with both the semantic priors of pretrained VLMs and the physical dynamics learned through future prediction. In practice, existing designs tend to erode the semantics of the pretrained backbone, suffer interference among heterogeneous objectives, and learn future prediction from scratch in pixel space, leaving the dynamics priors of pretrained video generators unexploited. We present InternVLA-A1.5, which builds the policy on a native VLM backbone that keeps training on VQA and subtask prediction, and attaches a lightweight unified expert for continuous action generation. Future prediction is recast as a latent-querying problem, where a small set of learnable foresight tokens condenses the task-relevant future into a compact latent code under the supervision of a frozen pretrained video generation model, so the policy inherits world-model dynamics priors without ever learning pixel-level generation. The video branch is discarded at inference, keeping real-time control. Pretrained on 1.2M robot episodes and 3M multimodal samples, InternVLA-A1.5 achieves the best overall results on all six simulation benchmarks. In the real world, the preserved semantics deliver the strongest compositional generalization on held-out instruction bindings, and the two designs together sustain long-horizon execution.