Zhejiang University and TeleAI Propose GR-BC for Precise Robotic Manipulation
Overview
A team from Zhejiang University, in collaboration with China Telecom AI Research Institute (TeleAI) and Gamma Intelligence, proposed Graph-Reweighted Behavior Cloning (GR-BC). The method uses graph-based trajectory analysis to identify effective action segments from imperfect demonstrations. Evaluated on four fine-grained tasks including DDR4 memory installation, GR-BC achieved an average success rate of approximately 91% using the X-VLA backbone.
Generated from attributed reports · Updated 1 hours ago
Event evidence and corrections
1 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Reported quantity · Reported success rate: 91 percent · reported trials · Differing source assertions
“图 5｜GR-BC 的离线迭代表现 实验结果：四项精细操作，平均成功率约 91% 真机实验覆盖四孔零件装配、显卡插装、双螺钉拧紧和内存条插装。”
Exact source · revision 1jiqizhixin
Reported quantity · Reported success rate: 98 percent · reported trials · Differing source assertions
“GR-BC 在上述四项任务上的最终成功率依次为 98%、93%、89% 和 83%， 平均约 91%；类 DAgger 等权聚合基线平均为 64%，相差约 27 个百分点。”
Exact source · revision 1jiqizhixin
Reported quantity · Reported success rate: 93 percent · reported trials · Differing source assertions
“GR-BC 在上述四项任务上的最终成功率依次为 98%、93%、89% 和 83%， 平均约 91%；类 DAgger 等权聚合基线平均为 64%，相差约 27 个百分点。”
Exact source · revision 1jiqizhixin
Reported quantity · Reported success rate: 89 percent · reported trials · Differing source assertions
“GR-BC 在上述四项任务上的最终成功率依次为 98%、93%、89% 和 83%， 平均约 91%；类 DAgger 等权聚合基线平均为 64%，相差约 27 个百分点。”
Exact source · revision 1jiqizhixin
Reported quantity · Reported success rate: 83 percent · reported trials · Differing source assertions
“GR-BC 在上述四项任务上的最终成功率依次为 98%、93%、89% 和 83%， 平均约 91%；类 DAgger 等权聚合基线平均为 64%，相差约 27 个百分点。”
Exact source · revision 1jiqizhixin
Reported quantity · evaluation trials per task: 30 other · reported trials
“每项任务固定评测 30 条成功、30 条失败轨迹。”
Exact source · revision 1jiqizhixin
Artifact availability · code: reported pending
“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。”
Exact source · revision 1jiqizhixin
Report timeline
Follow attributed reports and material updates.
- Machine Heart — Robotics on WeChatSignalZhejiang University and TeleAI Propose GR-BC for Precise Robotic Manipulation
A team from Zhejiang University, in collaboration with China Telecom AI Research Institute (TeleAI) and Gamma Intelligence, proposed Graph-Reweighted Behavior Cloning (GR-BC). The method uses graph-based trajectory analysis to identify effective action segments from imperfect demonstrations. Evaluated on four fine-grained tasks including DDR4 memory installation, GR-BC achieved an average success rate of approximately 91% using the X-VLA backbone.
Event coverage history
There is not enough continuous observation data to show a trend.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).