广告
加载中

字节跳动参投 自变量机器人获 10 亿元 A++ 轮融资|产业融资快报

李佳晅 2026-01-12 16:29
李佳晅 2026/01/12 16:29

邦小白快读

EN
全文速览

自变量机器人完成10亿元A++轮融资,获多家大厂投资,技术亮点突出。

1. 融资详情:近期完成10亿元A++轮融资,投资方包括字节跳动、红杉中国、北京信息产业发展基金等,是国内唯一同时被字节、美团、阿里投资的具身智能企业。

2. 公司背景:成立于2023年12月,创始人兼CEO王潜毕业于清华大学,是最早引入注意力机制的学者之一,自研主从遥操、外骨骼等数采设备。

3. 技术突破:软硬件全栈自研,发布“量子一号”“量子二号”机器人本体,核心零部件如机械臂、关节模组自研,大幅降低整机成本。

4. 应用领域:已进入工业制造、物流、养老等多个领域,WALL-A模型提升机器人零样本泛化能力。

5. 未来展望:2025年开源自研端到端具身基础模型WALL-OSS,推动技术普及,创始人强调下一阶段竞争是数据闭环和模型进化能力的竞争。

字节跳动等投资自变量机器人,显示品牌合作和产品研发趋势。

1. 品牌营销:投资方包括字节、美团、阿里,表明大厂对具身智能的重视,提供潜在品牌合作渠道。

2. 产品研发:自研“量子一号”“量子二号”机器人本体,核心零部件全面自研与算法适配,降低成本,启示品牌产品创新方向。

3. 消费趋势:进入养老、物流等领域,反映用户行为变化,如精细操作需求提升,WALL-A模型实现类人技能如手内重定向。

4. 用户行为观察:依托大规模真机强化学习,基础模型从交互中进化,解决长尾问题,显示消费者对智能化产品的接受度增强。

融资事件带来市场增长机会和技术可学习点。

1. 增长市场:自变量进入工业制造、物流、养老等新领域,提供消费需求变化下的商业拓展空间。

2. 机会提示:获字节、美团、阿里投资,显示合作方式可能,如平台招商或扶持政策;整机成本大幅下降,降低进入门槛。

3. 可学习点:自研端到端技术路线构建模型-真机进化闭环,WALL-A模型提升非结构化环境操作能力,卖家可借鉴应对事件风险。

4. 最新商业模式:开源WALL-OSS模型推动开放普及,提供正面影响,如降低技术采用风险,促进行业合作。

自变量机器人自研核心零部件启示生产数字化和商业机会。

1. 产品生产需求:自研机械臂、关节模组、动力驱动器等核心零部件,实现算法深度适配,大幅降低整机成本,提供设计优化方向。

2. 商业机会:进入工业制造领域,显示潜在合作需求,如零部件供应或定制生产;应用扩展到物流、养老,开拓新市场。

3. 推进数字化启示:大规模真机强化学习推进基础模型进化,解决长尾问题,工厂可借鉴数据采集设备(如主从遥操)提升自动化水平。

行业趋势聚焦新技术解决客户痛点和方案普及。

1. 行业发展趋势:具身智能快速发展,全球加快投入数据、模型、算力,WALL-A模型融合VLA与世界模型,推动多模态架构创新。

2. 新技术:首创VLA与世界模型深度融合,实现具身多模态思维链,提升时空预测和视觉因果推理能力。

3. 客户痛点:解决精细操作最后一厘米问题,如手内重定向,通过可学习记忆机制内化物理常识。

4. 解决方案:依托真机强化学习,模型自主进化;开源WALL-OSS模型促进技术开放,提供标准化服务方案。

投资和扩张显示平台需求和运营管理启示。

1. 商业对平台需求:字节、美团、阿里等投资,表明平台商对具身智能的看好,提供招商机会如合作开发。

2. 平台的最新做法:自变量进入工业、物流、养老多领域,启示平台运营管理扩展策略;模型自主解决长尾问题,帮助风险规避。

3. 平台招商:整机成本下降和开源模型(WALL-OSS)降低技术门槛,吸引更多合作伙伴。

4. 风向规避:端到端技术路线确保模型进化能力,减少执行风险,创始人观点强调数据闭环竞争,指导平台长期规划。

产业动向围绕技术创新和竞争焦点。

1. 产业新动向:具身智能进入数据闭环构建阶段,全球加速投入数据、模型、算力,自变量WALL-A模型实现世界模型机制融合。

2. 新问题:如何提升零样本泛化能力和精细操作,如攻克手内重定向技能;模型进化依赖真机交互学习。

3. 商业模式:开源WALL-OSS模型推动技术普及,提供开放研究平台;整机成本下降启示商业化路径。

4. 政策法规启示:创始人王潜观点强调基础模型与进化能力竞争,建议加强数据采集和模型迭代政策支持。

返回默认

声明:快读内容全程由AI生成,请注意甄别信息。如您发现问题,请发送邮件至 run@ebrun.com 。

我是 品牌商 卖家 工厂 服务商 平台商 研究者 帮我再读一遍。

Quick Summary

Embodied AI company Zibian Robotics has completed a 1 billion yuan A++ funding round with investments from major tech firms, highlighting significant technological advancements.

1. Funding Details: The recent 1 billion yuan A++ round was led by investors including ByteDance, Sequoia China, and Beijing Information Industry Development Fund. Zibian is the only embodied AI enterprise in China simultaneously backed by ByteDance, Meituan, and Alibaba.

2. Company Background: Founded in December 2023, CEO Wang Qian, a Tsinghua University graduate, is one of the earliest scholars to introduce attention mechanisms. The company has developed proprietary master-slave teleoperation systems and exoskeleton data collection devices.

3. Technological Breakthroughs: With full-stack in-house R&D in both hardware and software, Zibian launched the "Quantum One" and "Quantum Two" robot platforms. Key components like robotic arms and joint modules are self-developed, significantly reducing overall costs.

4. Application Areas: The robots are already deployed in industrial manufacturing, logistics, and elderly care. The WALL-A model enhances zero-shot generalization capabilities for robots.

5. Future Outlook: Zibian plans to open-source its end-to-end embodied foundation model, WALL-OSS, in 2025 to drive technology adoption. The founder emphasizes that future competition will center on data闭环 and model evolution capabilities.

Investments in Zibian Robotics by ByteDance and others signal trends in brand collaboration and product development.

1. Brand Marketing: Backing from ByteDance, Meituan, and Alibaba reflects major players' focus on embodied AI, offering potential partnership channels for brands.

2. Product R&D: The self-developed "Quantum" series robots, with fully customized core components and algorithm integration, reduce costs and inspire innovation in product design.

3. Consumer Trends: Expansion into elderly care and logistics indicates shifting user behaviors, such as rising demand for precision tasks. The WALL-A model enables human-like skills like in-hand reorientation.

4. User Behavior Insights: Leveraging large-scale real-world reinforcement learning, the foundation model evolves through interaction, addressing long-tail challenges and reflecting growing consumer acceptance of intelligent products.

Zibian's funding round presents market growth opportunities and actionable insights for sellers.

1. Market Expansion: Zibian's entry into industrial manufacturing, logistics, and elderly care opens new commercial avenues amid evolving consumer demands.

2. Opportunities: Investments from ByteDance, Meituan, and Alibaba suggest potential collaborations, such as platform partnerships or support policies. Lower robot costs reduce entry barriers.

3. Key Takeaways: The end-to-end self-developed technology creates a model-robot evolution loop. The WALL-A model improves performance in unstructured environments, offering sellers strategies to mitigate operational risks.

4. Emerging Models: Open-sourcing the WALL-OSS model promotes accessibility, reducing adoption risks and fostering industry cooperation.

Zibian's in-house core component development offers insights into production digitization and business opportunities.

1. Production Needs: Self-developed robotic arms, joint modules, and power drivers enable deep algorithm integration, cutting costs and informing design optimizations.

2. Business Opportunities: Industrial manufacturing applications reveal potential for part supply or custom production. Expansion into logistics and elderly care unlocks new markets.

3. Digitization Insights: Large-scale real-world reinforcement learning drives model evolution, solving long-tail issues. Factories can adopt data collection tools like master-slave teleoperation to enhance automation.

Industry trends highlight new technologies addressing client pain points and solution scalability.

1. Sector Growth: Embodied AI is advancing rapidly, with global investments in data, models, and computing. Zibian's WALL-A model integrates VLA and world models, spurring multimodal architecture innovation.

2. Technological Innovations: The pioneering fusion of VLA and world models enables embodied multimodal reasoning, improving spatiotemporal prediction and visual causal inference.

3. Client Pain Points: Solutions address fine-motor challenges like in-hand reorientation, internalizing physical intuition via learnable memory mechanisms.

4. Service Solutions: Real-world reinforcement learning allows autonomous model evolution. Open-sourcing WALL-OSS promotes standardized service offerings.

Investments and expansion reveal platform demands and operational insights.

1. Platform Needs: Backing from ByteDance, Meituan, and Alibaba underscores platform interest in embodied AI, creating partnership opportunities like co-development.

2. Platform Strategies: Zibian's multi-sector presence informs platform expansion tactics. Autonomous model evolution helps mitigate risks.

3. Partnership Opportunities: Lower robot costs and open-source WALL-OSS reduce technical barriers, attracting collaborators.

4. Risk Management: End-to-end technology ensures model adaptability, minimizing execution risks. The founder's focus on data闭环 competition guides long-term planning.

Industry developments center on technological innovation and competitive dynamics.

1. Trends: Embodied AI is entering a data闭环 phase, with global acceleration in data, model, and compute investments. Zibian's WALL-A model achieves world model integration.

2. Research Challenges: Enhancing zero-shot generalization and fine-motor skills (e.g., in-hand reorientation) remains key. Model evolution relies on real-robot interaction.

3. Commercial Models: Open-sourcing WALL-OSS fosters technology diffusion, offering research platforms. Cost reductions illuminate commercialization pathways.

4. Policy Implications: Founder Wang Qian emphasizes competition in foundation models and evolution capabilities, advocating policy support for data collection and iterative model development.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

【亿邦原创】1月12日,自变量机器人宣布于近期完成10亿元A++ 轮融资。本轮融资由字节跳动、红杉中国、北京信息产业发展基金、深创投、南山战新投、锡创投等投资机构及多元地方平台联合投资。值得关注的是,除字节外,自变量此前也曾先后获得美团、阿里的投资,是国内唯一同时被这三家互联网大厂投资的具身智能企业。

自变量机器人成立于2023年12月。公司创始人兼CEO为王潜,毕业于清华大学,是最早在神经网络中引入注意力机制的学者之一。作为国内最早规模化扩展真机数据采集的公司,自变量自研了主从遥操、外骨骼、无本体等多种数采设备,实现了各种数采设备上的数据验证和模型突破。

自变量机器人坚持软硬件全栈自研,设计发布了“量子一号”“量子二号”两款高性能的机器人本体,同步实现了机械臂、关节模组、动力驱动器、主控制器等核心零部件的全面自研与算法深度适配,促成整机成本的大幅下降。目前,自变量机器人已逐步进入工业制造、物流、养老等多个领域。

自变量自研的WALL-A模型,核心架构首创VLA与世界模型深度融合的系统范式。作为原生的多模态输入输出架构,WALL-A率先实现了具身多模态思维链。WALL-A利用世界模型机制进行时空状态预测,协同视觉因果推理理解环境反馈,并通过可学习记忆机制从数据中内化物理常识,显著提升了机器人执行非结构化环境中移动操作任务的零样本泛化能力。

同时,依托于大规模真机强化学习,基础模型进一步在与真实物理世界的交互中获得高质量学习经验,自主解决长尾问题,实现机器人能力的持续进化。自变量以完全端到端技术路线构建了物理世界基础模型-真机自主进化的技术闭环。

自变量基础模型的进化还解锁了高自由度灵巧手的潜力,机器人自主掌握了手内重定向等类人技能——从使用工具,到发牌这类对指尖力控要求极高的精细动作,成功攻克了具身智能精细操作的最后一厘米。2025年9月,自变量还开源了其自研端到端具身基础模型WALL-OSS,推动具身智能技术的开放普及。

自变量机器人创始人兼CEO王潜表示:“具身智能的下一阶段竞争,本质上还是数据闭环构建的基础模型与模型进化能力的竞争”。在这个判断下,全球正在从数据、模型、算力等多个方面加快投入,快速推进具身智能的发展。

文章来源:亿邦动力

广告
微信
朋友圈

这么好看,分享一下?

朋友圈 分享

APP内打开

+1
+1
微信好友 朋友圈 新浪微博 QQ空间
关闭
收藏成功
发送
/140 0