English(EN) HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning

机器人学习系统HIDRA使用双路由注意力实现持续适应

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-28 04:00

研究人员开发了HIDRA，一个用于机器人领域终身模仿学习的新框架。该系统使用分层双注意力机制，能够从多模态演示中可靠地选择和优化专家动作，而无需存储或重新访问过去的数据。HIDRA还结合了几何正则化，以在表示漂移的情况下稳定专家路由，从而实现持续适应和可扩展的专家集。在LIBERO基准套件上的评估表明，HIDRA的性能优于无重放和有重放的方法，尤其是在高干扰和不同语言指令的条件下。 AI

影响增强了机器人代理在不忘记过去任务的情况下从新数据中持续学习的能力。

排序理由详细介绍机器人领域终身模仿学习新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 13 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 详细介绍机器人领域终身模仿学习新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, other; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Fanqi Yu, Matteo Tiezzi, Cigdem Beyan, Tommaso Apicella, Vittorio Murino · 2026-10-28 04:00

HIDRA：用于无重放终身模仿学习的分层双路由注意力

arXiv:2610.36374v1 Announce Type: cross Abstract: Robotic agents operating in real-world environments must adapt continuously from a stream of multimodal demonstrations, often under constraints that preclude storing or revisiting past data. In this replay-free, single-pass lifelo…

本文如何被排名

报道来源 [1]

HIDRA：用于无重放终身模仿学习的分层双路由注意力

相关实体

相关话题