Agent 走向具身世界：从语言智能到机器人「大脑指挥官」
Ranking
Observed public metrics from 1 member.
Merged summary
TL;DR — 这些来源共同展示了 Agent 从语言智能走向现实执行系统的趋势：通过任务分解、上下文理解、工具或策略调度与反馈闭环，Agent 正成为机器人和企业工作流的“指挥中枢”。RoboHarness 聚焦异构机器人策略的协同，Doubao Work 则聚焦飞书组织上下文中的自主办公。
- 华为诺亚方舟实验室的 RoboHarness 使用编码 Agent 编排 VLA、世界-动作模型、强化学习及任务与运动规划等异构策略，无需重新训练或修改底层策略即可执行零样本、长时程机器人任务。
- RoboHarness 根据场景量化信号、执行历史和在线反馈拆解任务并动态路由子任务；其 Memory Bridge 会检索成功轨迹，规划可靠的策略交接状态。移除该模块后，完整任务成功率从 86.0% 降至 60.4%。
- 字节跳动的 Doubao Work 可自主拆解办公任务，并操作文件、浏览器、云电脑和内容生成工具，产出分析、仪表盘、图片、视频及网页等结果。
- Doubao Work 通过飞书账号继承组织身份与权限，可在授权范围内调用聊天、文档、会议、任务、审批和多维表格，并将讨论或业务数据转化为可协作、可回流工作流的成果。
- 两者都强调可靠执行：RoboHarness 通过历史轨迹与状态转换降低机器人策略交接风险；Doubao Work 则保留采购信息的不确定性，将无法核实的价格、库存和交期标记为估算或待询价。
注： 两个来源实际强调的是不同系统——雷峰网聚焦机器人策略编排，量子位聚焦企业办公 Agent；它们的共同点是以 Agent 作为复杂执行环境中的上层协调者。
Sources (2)
Agent 走向具身世界：从语言智能到机器人「大脑指挥官」
TL;DR - Huawei Noah’s Ark Lab’s RoboHarness uses a coding agent to orchestrate heterogeneous robot policies—VLA, world-action models, reinforcement learning, and task-and-motion planning—for zero-shot, long-horizon tasks. It matters because it improves robotic capability through modular coordination without retraining or modifying the underlying policies.
- RoboHarness decomposes tasks and dynamically routes subtasks using quantitative scene signals, execution history, and online feedback.
- Its Memory Bridge retrieves successful trajectories and plans transitions into states where the next policy can reliably take control.
- Removing Memory Bridge reduced full-task success from 86.0% to 60.4%, highlighting policy handoffs as a key failure point.
- The approach complements embodied foundation-model scaling with an agentic hierarchy, though it remains limited by available policies and historical execution data.
深度实测「豆包工作」+飞书：目前最接近企业Agent终局的答案
TL;DR - ByteDance launched Doubao Work, a productivity agent deeply integrated with Feishu’s organizational identity, permissions, communications, and workflows. The integration highlights enterprise context—not generic tool use—as a key differentiator for workplace agents.
- Doubao Work autonomously decomposes tasks and operates files, browsers, cloud computers, and content-generation tools to produce images, videos, webpages, analyses, and dashboards.
- Feishu account login lets the agent access authorized chats, documents, meetings, tasks, approvals, and multidimensional spreadsheets without repeated uploads or separate identity configuration.
- In the reported tests, it converted editorial discussions into a collaborative topic tracker and analyzed project, sales, and operational data before returning outputs to Feishu workflows.
- The product preserved uncertainty in procurement research by marking unverifiable prices, inventory, and delivery dates as estimates or pending quotation rather than filling gaps.