研究人员正在开发新方法来提高机器人操作能力，特别是在涉及多个顺序动作的长时域任务方面。一种名为ARLI的方法，通过引入状态增强来恢复对强化学习至关重要的马尔可夫假设，从而解决了大型通用机器人策略中的推理延迟挑战。另一种名为DELE-w0.5的方法，绕过了视频生成，直接从预测的未来状态推断动作，降低了训练成本和推理延迟。第三个系统BATON通过将子任务视为探索单元并采用感知转换的记忆来管理任务执行和适应，来应对长时域操作。 AI
影响 这些机器人操作方面的进步可能导致在复杂、现实环境中的机器人系统更具能力和效率。
排序理由 该集群包含三篇详细介绍机器人操作新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2608.25757v1 Announce Type: cross Abstract: Generalist vision--language--action (VLA) policies learn long-horizon behavior mainly through short-horizon action prediction and reveal little beyond sampled commands. This creates two coupled bottlenecks: a single action target …
arXiv:2608.23831v1 Announce Type: cross Abstract: While reinforcement learning (RL) allows generalist robot policies to continually improve during deployment, the large model size of modern generalist policies, such as VLAs, poses a fundamental obstacle to effective RL improvemen…
arXiv:2608.22067v1 Announce Type: cross Abstract: World-Action Models (WAMs) build robot control on video-generation backbones, which jointly predict dense future visual trajectories and robot actions. We argue that video generation is an unnecessary intermediate objective for wo…
Long-horizon robot manipulation chains many contact-rich skills into one multi-stage task. Vision-language-action (VLA) models increasingly master the individual skills, yet the chain still fails: errors compound beyond the policy's ability to correct, and one subtask silently co…