具身智能的价值不仅在于重复执行单一动作，更在于能够在真实环境中灵活应对不同任务。要形成这种泛化能力，需要高质量、标准化且规模足够大的训练数据。福莱新材基于柔性触觉传感器和灵巧手方案交付经验，正在将触觉能力延伸至具身智能数据采集终端。
上海证券报报道，福莱新材首席科学家陈书厅在接受采访时，介绍了公司在数据采集终端核心技术、客户覆盖和产能支撑等方面的能力储备，并表示数据采集终端有望成为公司触觉传感业务新的应用方向。
具身智能训练数据缺口或达上亿小时
机器人模型要具备更强的泛化能力，需要在大量真实任务中积累训练数据。行业目前已有的数据规模与未来需求仍存在明显差距，因此，大规模、多样化的数据采集正在成为具身智能产业的重要基础环节。
在拿取水杯、拧螺丝等精细操作中，机器人不仅需要看见物体，还需要感知接触位置、握持力度和滑移状态。触觉数据可以同时记录成功操作和失败操作，让模型从不同结果中学习如何稳定抓取、调整力度并完成动作。
将灵巧手触觉经验迁移至数采终端
福莱新材将自身定位为数据采集终端提供方，重点提供触觉采集硬件和数据能力，不直接从事模型训练。根据公开报道，公司当前以数采手套作为产品形态，并划分为通用版和专业版。
通用版面向人在自然操作过程中的数据采集，使操作者能够在工厂、实验室、农田等场景中完成任务，同时记录指尖和手掌区域的受力信息。专业版则根据特定灵巧手厂商、数据格式和训练任务进行定制。
数采手套的技术难点之一，是将安装在机器人刚性结构上的触觉传感能力迁移到柔软、易形变的手套基材。手套受到外力时容易发生形变，力的传递路径也更复杂，需要解决传感器固定、信号稳定、多区域同步采集和穿戴舒适性等问题。
客户覆盖与交付能力支撑产品推进
报道提到，福莱新材的柔性触觉传感器已服务国内外多类机器人及灵巧手客户。公司已有的触觉产品交付经验、研发团队和生产能力，为数据采集终端的开发、客户适配和后续规模化交付提供基础。
完整的数据采集方案并不只有传感器，还包括采集电路、通信接口、标定方法、数据处理和软件工具。只有统一采样时间、数据格式与坐标定义，触觉信息才能与视觉、关节位置和动作轨迹等数据共同用于模型训练。
多场景推动触觉传感器应用落地
除具身智能数据采集外，福莱新材还在机器人触觉、工业检测、新能源和智能交互等方向推进柔性传感技术应用。不同场景对传感器结构、量程、接口和算法提出不同要求，需要结合客户产品持续进行验证和适配。
数据采集终端及柔性触觉传感业务仍处于持续开发和商业化推进阶段，具体技术指标、客户合作和经营数据应以公司正式公告及产品资料为准。
The value of embodied intelligence lies not only in repeating a single action, but in adapting flexibly to different tasks in real environments. Developing this generalization capability requires high-quality, standardized training data at sufficient scale. Drawing on its flexible tactile sensors and dexterous-hand solution delivery, Fulai is extending tactile capabilities to embodied-AI data acquisition terminals.
Shanghai Securities News reported that Fulai Chief Scientist Shuting Chen described the company\'s capabilities in core data-acquisition-terminal technology, customer coverage and production support. He said data acquisition terminals may become a new application direction for Fulai\'s tactile sensing business.
The training-data gap may reach hundreds of millions of hours
Robotic models need experience from a large number of real tasks to develop stronger generalization. A substantial gap remains between the data currently available to the industry and future demand, making large-scale, diverse acquisition a critical part of the embodied-intelligence ecosystem.
In fine tasks such as picking up a cup or tightening a screw, a robot must not only see the object but also sense contact location, grip force and slip. Tactile data records both successful and failed operations so models can learn how to grasp reliably, adjust force and complete an action.
Transferring dexterous-hand experience to data acquisition terminals
Fulai positions itself as a provider of data acquisition terminals, concentrating on tactile acquisition hardware and data capabilities rather than model training. According to public reporting, the current product takes the form of a data-acquisition glove in general-purpose and professional versions.
The general-purpose version records natural human operations in factories, laboratories, farms and other environments while capturing force information from fingertips and palms. The professional version is customized for a specific dexterous-hand manufacturer, data format and training task.
A key technical challenge is transferring tactile sensing from a rigid robotic structure to a soft, deformable glove substrate. External force deforms the glove and complicates load paths, requiring solutions for sensor fastening, signal stability, synchronized multi-region acquisition and wearing comfort.
Customer coverage and delivery capabilities support development
The report states that Fulai flexible tactile sensors already serve multiple robotic and dexterous-hand customers in China and overseas. Existing delivery experience, R&D resources and manufacturing capacity provide a foundation for terminal development, customer adaptation and future scaled delivery.
A complete acquisition solution includes more than sensors: acquisition electronics, communication interfaces, calibration, data processing and software tools are also required. Tactile information can be used together with vision, joint position and motion trajectories only when sampling time, formats and coordinate definitions are unified.
Multiple applications advance tactile sensing deployment
Beyond embodied-AI data acquisition, Fulai is advancing flexible sensing in robotic touch, industrial inspection, new energy and intelligent interaction. Each application has different requirements for structure, range, interface and algorithms and therefore requires continuous customer validation and adaptation.
Data acquisition terminals and flexible tactile sensing remain under continued development and commercialization. Technical indicators, customer cooperation and business data should be confirmed against official company announcements and product documents.
