Solution 01 / Flagship

科学实验机器人与自主实验室 Scientific robots and autonomous labs

让实验机器人理解科学任务、执行复杂操作，并在实验反馈中持续进化。 Enable lab robots to understand scientific tasks, execute complex operations, and evolve through experiment feedback.

AI4S Robot Scientist Autonomous Lab Flagship
Who it serves

为科学实验的前沿探索者构建 Built for the frontiers of scientific experimentation

AI Lab Solution 面向需要将机器人从自动化执行设备升级为协作科学智能体的各类实验场景。 For labs that need to upgrade robots from automation devices to collaborative scientific agents.

高校实验室 University labs

跨学科研究中引入具身智能，降低实验机器人使用门槛。 Introduce embodied intelligence into interdisciplinary research, lowering the barrier for lab robotics.

科研院所 Research institutes

在材料、化学、生物等领域实现长程实验自动化与数据闭环。 Automate long-horizon experiments and close the data loop in materials, chemistry, and biology.

企业研发中心 Corporate R&D

加速配方筛选、工艺优化和质检流程，沉淀研发知识资产。 Accelerate formulation screening, process optimization, and QA workflows; build R&D knowledge assets.

医院与医学实验室 Hospital & medical labs

在样本处理、药物筛选和高通量实验中实现标准化与可追溯。 Standardize and trace sample handling, drug screening, and high-throughput experiments.

AI4S 平台机构 AI4S platforms

将具身操作能力接入 AI for Science 基础设施，补齐物理实验环节。 Connect embodied operation capabilities to AI4S infrastructure, closing the physical experiment gap.

自动化实验室建设方 Lab automation builders

从定制化自动化走向智能化自主实验，降低系统集成和维护成本。 Move from custom automation to intelligent autonomous experimentation, reducing integration and maintenance costs.

六类典型自主任务 Six typical autonomous tasks

从样品操作到技能沉淀，AI Lab Solution 将"实验需求 → 任务拆解 → 执行 → 观察 → 纠偏 → 学习"形成完整闭环。 From sample handling to skill accumulation, AI Lab Solution closes the full loop: experiment need → task decomposition → execution → observation → correction → learning.

样品搬运与上下料 Sample transfer & loading

样品瓶、微孔板、试管架的抓取、搬运与自动化上下料 Grasp, transfer and automated loading of vials, microplates and tube racks

仪器操作与分拣 Instrument & sorting

实验仪器开关、旋钮、按钮和触屏界面操作，器具智能分拣 Operate instrument switches, knobs, buttons and touchscreens; intelligent tool sorting

试剂与耗材管理 Reagent & consumables

试剂分装、耗材补给、废液处理，实验物料全流程追踪 Reagent dispensing, consumable replenishment, waste handling, full material traceability

观察记录与状态确认 Observation & confirmation

实验过程中的视觉观察、读数记录和阶段状态自动确认 Visual observation, reading capture and automated stage confirmation during experiments

异常处理与人类纠偏 Exception & correction

异常状态自动检测、安全暂停，通过人机交互完成纠偏并继续 Auto-detect anomalies, safe pause, complete correction via human interaction and resume

技能数据沉淀与复用 Skill accumulation

实验操作技能数据采集、模型训练与跨实验台跨场景的复用 Skill data capture, model training and cross-workstation cross-scenario reuse

Demo

自主任务执行 Autonomous task execution

从任务输入到动作执行，AI Lab Solution 在真实实验场景中的全链路演示。 End-to-end demonstrations of AI Lab Solution in real experimental scenarios, from task input to action execution.

Solution architecture

四项内部技术与 Dora 底座 Four internal technologies and the Dora foundation

四项 Oasis 内部技术在 AI Lab 研究中承担不同角色，通过 Dora 底座协同，形成从数据采集到模型部署的完整链路。 Each internal Oasis technology has a role in AI Lab research, coordinated through the Dora foundation to form a complete chain from data capture to model deployment.

Oasis WAM — 具身大脑 Oasis WAM — Embodied brain 嵌入实验机器人，负责任务理解、长程规划和精确动作生成，在安全约束和人类引导下持续优化操作策略。 Embedded in lab robots for task understanding, long-horizon planning, and precise action generation, with safety constraints and human guidance.
Oasis Trace — 轨迹采集 Oasis Trace — Trajectory capture 采集实验员操作、机器人操作、第一人称视角和遥操作轨迹，为技能学习和数据治理提供高质量真实输入。 Capture human operator actions, robot operations, first-person views, and teleoperation trajectories for skill learning and data governance.
Oasis Data — 数据治理 Oasis Data — Data governance 管理实验操作数据、轨迹数据、标注数据、失败样本和训练集，支撑模型训练、评测和版本迭代。 Manage experiment data, trajectories, annotations, failure samples, and training sets for model training, evaluation, and versioning.
OASIS SIM — 具身仿真与数据合成 — 仿真验证 OASIS SIM — Embodied Simulation & Synthetic Data — Simulation & validation 构建实验室虚拟场景，生成长尾操作数据，在真实部署前验证策略安全性，降低物理试错风险。 Build virtual lab scenes, generate long-tail operational data, validate policy safety before deployment, reducing physical trial risks.
Dora — 开源底座 Dora — Open-source foundation 提供低延迟多模态数据流、机器人控制和边缘推理 Runtime，是整个方案的系统级支撑层。 Provide low-latency multimodal dataflow, robot control, and edge inference runtime as the system-level foundation.

将你的实验室带入具身智能时代 Bring your lab into the embodied intelligence era

无论是单台实验机器人还是整建制自主实验室，欢迎联系我们讨论具体实验任务和实施条件。 Whether you are considering a single laboratory robot or a complete autonomous laboratory, contact us to discuss specific experimental tasks and implementation requirements.

联系合作 Talk to us