Zhejiang University and TeleAI Propose GR-BC for Robotic Manipulation, Accepted at CoRL 2026
Zhejiang University and TeleAI proposed Graph-Reweighted Behavior Cloning (GR-BC), a method that builds a graph from multiple robot attempts and uses graph search to identify effective actions for precise manipulation. The work was accepted at CoRL 2026, with paper, code, and project page publicly available.
Coverage timeline
机器之心机器之心
用螺丝刀拧紧螺丝，听起来并不复杂。 但真正动手时，往往还要挪一点、转一点、退出来，再重新对齐。换成遥操作机器人采集数据，这个过程只会更考验耐心。即使是 “大国工匠” 般熟练的数采员，面对精细度要求极高的任务，也未必能一次交出从头到尾都足够丝滑的示范。一条最终成功的轨迹里，可能包含多次无效调整；一条最终因超时或碰撞而失败的尝试中，却也可能有几步对齐做得很好。 这给视觉语言动作模型的后训练提出了一个问题： 如果数采员自己都很难采出一条近乎完美的轨迹，机器人应该向什么样的经验学习？ 浙江大学沈春华团队联合中国电信人工智能研究院（TeleAI）& 伽马智能白辰甲团队， 围绕这个问题提出了 Graph-Reweighted Behavior Cloning（GR-BC）。 团队考虑的是另一种可能：一条更好的成功路径，未必完整存在于某一次示范中，却可能隐藏在多次尝试之间的连接关系里。把这些经验放进同一张图， 能否先在图的 “脑海” 里找到一条更好的路，再由图担任裁判，指导机器人学习值得模仿的动作？ 目前， 该工作已被 CORL 2026 接收，论文与代码已公开。 论文标题：Graph-Reweighted Behavior Cloning with Human Interventions for Precise Robotic Manipulation 论文链接：https://arxiv.org/pdf/2610.20504 项目主页：https://gr-bc.github.io 代码链接：https://github.com/Gamma-Robot/GR-BC 核心方法：用图检索进度，再为动作加权 图 1｜GR-BC：构图、反向搜索、加权训练与新轨迹回流。 第一步：把分散经验连接起来 要评判一个动作好不好，首先需要知道，机器人当前距离任务完成还有多远。团队观察发现： 精细任务中的进度，具有很强的任务内相对性。 把同一任务的多次尝试放在一起，利用它们之间的联系，可以为进度判断提供更具体的依据。 GR-BC 使用冻结的 MCR（manipulation-centric representation）视觉编码器提取三个相机视角的表征，与夹抓状态结合，共同编码图中节点。 边有两种。 时间边 连接同一轨迹上按固定间隔抽样的相邻的节点，代价等于跨越的帧数，因此图上的距离天然对应真实的
Related stories
UCSB and Microsoft Propose ScholarEvolve for AI Agents to Read Papers and Self-Modify Code
Researchers from UC Santa Barbara and Microsoft introduced ScholarEvolve, a method that automates the research process of reading literature, implementing algorithms, and evaluating results to enable AI agents to evolve their harness code without modifying underlying model parameters. The approach addresses the limitation of prior coding-agent-based harness optimization, which relied on the agent's existing knowledge and observed failures, by allowing agents to draw on the latest academic findings. The paper and code are publicly available.
2 reports · 1 sources
Fudan, Tencent Hunyuan, Tsinghua Release ExplorationBench to Test AI Rule Discovery
Fudan University, Tencent Hunyuan, and Tsinghua University released ExplorationBench, a benchmark measuring how AI systems discover rules in unfamiliar environments. It includes two altered 'alien worlds'—the programming language AlienCode and the logic proof system AlienLogic—where models must learn rules through trial and error. Results showed that without environment feedback, scores remained low (max 11.0%), while autonomous exploration led to high scores (up to 89.0%).
Google Researchers Introduce T2J-Bench for PyTorch-to-JAX Migration
Google researchers and collaborators introduced T2J-Bench, a benchmark of 63 repository-migration tasks for evaluating coding agents on execution-grounded PyTorch-to-JAX migration, along with the TwinHarness method. In tests across 20 agent-model configurations, the strongest baseline scored 0.676; TwinHarness improved Claude-Code with opus-5 from 0.649 to 0.723, a relative gain of 11.4%.
OpenAI, Google, Microsoft, Anthropic back shared robots.txt AI opt-out standard
OpenAI, Google, Microsoft, and Anthropic announced they will adopt a new IETF standard that lets publishers opt out of AI scraping across multiple crawlers via a shared policy token in robots.txt. The token, proposed by Anthropic technical lead Kevin Kelley, allows website operators to express AI training preferences once instead of setting them per company. This move aims to simplify publisher control over AI data usage.
3 reports · 3 sources
News updates every hour. Subscribe in your RSS reader to get new stories automatically.
Subscribe