XINMING 合肥芯明智能科技有限公司
Hefei chip designer founded 2020; NU4000/NU4100 are the only mass-produced single-chip SoCs integrating 3D stereo vision, AI, and SLAM for robotics.
Hefei Xinming Intelligent Technology Co., Ltd., operating under the brand XINMING, is a high-tech enterprise founded in 2020 and headquartered in Hefei, Anhui Province, China, with subsidiaries and branches in Shanghai, Beijing, Shenzhen, and Israel. The company specializes in the design of spatial computing and artificial intelligence system-on-chip (SoC) solutions, with a stated mission to make spatial intelligence ubiquitous through continuous innovation. XINMING's primary products are its NU-series SoC chips and associated vision modules.
The NU4000 is a 12nm process system-level chip integrating 3D depth imaging, AI processing, and SLAM (Simultaneous Localization and Mapping) for spatial computing applications, and is described as the world's only mass-produced single-chip that integrates chip-based real-time 3D stereo visual perception, AI inference, and SLAM in a single device. The NU4100 is a successor 12nm SoC with enhanced AI and 3D imaging capabilities, supporting high-resolution imaging and SLAM for spatial intelligence devices. The NU4500 has been announced as forthcoming. The module lineup includes C158, R112, R130, R132, R158, R216g, R232, and R258, covering USB and industrial Ethernet interface variants for integration into robotic systems, drones, XR devices, and consumer electronics. Solution verticals on the company's website span service robots, logistics robots, industrial robots, humanoid robot perception (including dexterous hand near-field high-precision sensing), AIoT applications in agriculture and livestock, logistics drones, AR/MR and metaverse, 3D scanning for medical imaging, and automotive low-speed assisted driving.
The company has raised a Series A+ round led by Kaiyuan Industrial, with additional investors including Hefei Industry Investment Group and Feixi Industrial Investment. In April 2025 the company completed its Series A+ financing round of hundreds of millions of yuan.
XINMING debuted at Electronica China and the 2026 Zhangjiang Embodied Intelligence Developer Conference, and exhibited at the 28th Beijing Science and Technology Exhibition in May 2026. The R216g 3D visual AI module was officially released in June 2026, achieving breakthroughs in perception accuracy, processing efficiency, and industrial reliability. The company's contact phone is +86 021-61652308 and general inquiries email is [email protected].
Primary type & automation activities this supplier delivers:
NU4000 Spatial Intelligence SoC
12nm single-chip SoC integrating real-time 3D stereo vision, AI inference, and SLAM—world's only mass-produced chip combining all three in a single device.
NU4100 Spatial Intelligence SoC
Enhanced 12nm SoC with upgraded AI and high-resolution 3D imaging plus SLAM, targeting spatial intelligence devices including humanoid robots and XR.
R216g 3D Vision AI Module
- Model: R216g
- Launch Year: 2026
Industrial-grade 3D visual AI module achieving triple breakthroughs in perception accuracy, processing efficiency, and industrial reliability; released June 2026.
Contact XINMING
WEBSITE
https://www.xinming-ai.comPHONE
+86 021 61652308HEADQUARTERS
China
Company Facts
Founded
2020
Primary Role
Component Supplier
Company Size
-
Primary Region
Asia-Pacific
Annual Sales
-
Funding Stage
Early-Stage VC
Funding Total
-
Keywords
- XINMING
- 合肥芯明智能科技有限公司
- AI chip
- 3D vision
- spatial intelligence
- SoC
- robot perception
- spatial computing chip
- NU4000
- NU4100
- 3D vision AI
- SLAM chip
- stereo vision module
- embodied intelligence
- humanoid robot perception
- Component Supplier
- Perception & Vision
- Navigation & Positioning
- Manufacturing
- Logistics & Supply Chain
IROS 2026 Category Award Finalists: All 22 Papers
From humanoid tennis to bin picking and farm robots: the 22 finalists for IROS 2026's eight category awards, and what each one means for industry.
IROS 2026 by the Numbers: Papers, Countries, Topics
RobotToday analyzed all 1,933 IROS 2026 papers: acceptance fell to 36.5%, the U.S. overtook China, and VLA titles jumped from 4 to 65.
IROS 2026 Best Paper Finalists: South Korea Leads
Ten papers made the IROS 2026 Best Paper and Best Student Paper shortlist, four from South Korea. What each one does and what it means for industry.
New 'SANDO' System Keeps UAVs Safe from Collisions in Wildfire Areas
The 'SANDO' system enhances the safety of uncrewed aerial vehicles (UAVs) by ensuring their flight paths remain collision-free, particularly in hazardous environments like wildfires. This technology is crucial as UAVs are often deployed to gather vital information in such dangerous situations, where they face risks from sudden flare-ups and falling debris. The importance of the 'SANDO' system lies in its ability to protect UAVs from potential damage during operations in uncharted territories. By avoiding collisions with obstacles such as tree branches, the system allows for safer data collection, which is essential for rescuers responding to wildfires. This advancement could significantly improve the effectiveness of UAVs in emergency response scenarios. Looking ahead, the implementation of the 'SANDO' system could lead to broader applications of UAVs in various challenging environments. As the technology develops, stakeholders will be keen to observe its impact on UAV operational safety and efficiency in disaster response efforts. No further timeline was disclosed at the time of publication.
Robotics
MIT Develops SANDO: A Collision-Free Navigation System for UAVs in Unknown Environments
MIT researchers have developed an autonomous navigation system named SANDO, designed for uncrewed aerial vehicles (UAVs) to navigate safely in unpredictable environments. This system ensures that UAVs can avoid collisions with unknown moving obstacles, which is crucial in high-risk scenarios like search and rescue missions or disaster response. The significance of SANDO lies in its ability to provide formal safety guarantees in dynamic environments, where traditional navigation systems fall short. Unlike existing planners that only work in static settings or lack formal guarantees, SANDO uses a mathematical approach to ensure collision avoidance, making it particularly valuable for applications requiring real-time adaptability. Looking ahead, the potential applications for SANDO are vast, including emergency medical supply deliveries and exploration in hazardous areas. As the technology matures, it could revolutionize how UAVs operate in complex environments, enhancing safety and efficiency. No further timeline was disclosed at the time of publication.
Research Computer science and technology Algorithms Drones Robotics Autonomous vehicles
TwelveLabs Launches Pegasus 1.6 for Enhanced Video Understanding in Physical AI
TwelveLabs Inc. has unveiled its Pegasus 1.6 model, designed to enhance video understanding for physical AI applications. This model addresses the challenge of transforming complex video footage into actionable insights, enabling robotics and AI teams to train on real human experiences rather than starting from scratch. The significance of Pegasus 1.6 lies in its ability to provide temporal context, spatial reasoning, and task completion judgment, which are crucial for machines to navigate real-world environments. According to Jae Lee, co-founder and CEO of TwelveLabs, the model allows for a more precise understanding of actions and interactions captured in egocentric video, facilitating the development of smarter robotics systems. Looking ahead, TwelveLabs aims to further refine its video intelligence platform, focusing on the video understanding layer to improve training data for robotics teams. The company emphasizes that Pegasus 1.6 can utilize existing video data and analyze still images, making it a versatile tool for various applications in physical AI. No further timeline was disclosed at the time of publication.
Artificial Intelligence Artificial Intelligence / Cognition Assembly Human Robot Interaction / Haptics Manufacturing Microprocessors / SoC
Teledyne FLIR Introduces AI-Driven Air-to-Ground Tracking Software for UAS
Teledyne FLIR OEM has unveiled Prism A-ISR, a cutting-edge software module aimed at enhancing air-to-ground detection and tracking for uncrewed aircraft systems (UAS) and airborne intelligence, surveillance, and reconnaissance (ISR) platforms. Announced on October 6, this software leverages artificial intelligence to detect, classify, and track multiple ground objects in real time, utilizing data from various sensor types including midwave infrared (MWIR) and visible electro-optical (EO) sensors. The significance of Prism A-ISR lies in its ability to perform edge processing, enabling airborne systems to analyze sensor data independently of remote computing resources. This capability is crucial for applications requiring persistent surveillance and defense missions, as it enhances long-range situational awareness while minimizing false alarms. Teledyne FLIR's advanced image processing tools, including patented denoise and super-resolution technology, further improve detection capabilities in challenging conditions. Looking ahead, Teledyne FLIR's AIMMGen synthetic model-training service allows developers to quickly adapt the software for new target classes, streamlining the creation of mission models. The software is compatible with NVIDIA and Qualcomm embedded AI platforms and several Teledyne FLIR camera modules, including the NDAA-compliant Boson+ and Neutrino thermal camera families. No further timeline was disclosed at the time of publication.
AI Applications C-UAS Defense defense Drone News
Overview AI Introduces OV Spark and OV Spark Pro AI Inspection Cameras
Overview AI has launched the OV Spark and OV Spark Pro, a new line of AI inspection cameras featuring an agentic AI layer that communicates in plain English. The cameras come equipped with 10 inspection tools, including seven deep learning AI models and three ready-to-use tools for barcodes, measurements, and color matching. The introduction of the OV Spark cameras is significant as they address the challenges of consistency in inspection processes. Christine Chacko, a technical product marketing manager at Overview AI, highlighted that the cameras adapt to the shifting and rotating parts in a robot cell, ensuring accurate inspections. This innovation aims to enhance defect detection across factory lines. Looking ahead, Overview AI's mission is to ensure that every defect is caught in every factory, with the OV Spark designed to minimize setup work. The cameras are currently available for purchase directly from Overview AI and through distribution partners. No further timeline was disclosed at the time of publication.
Artificial Intelligence Artificial Intelligence / Cognition Cameras / Imaging / Vision Design / Development Manufacturing News
Epson Showcases Color Labeling and Robotic Automation at PACK EXPO 2026
Epson has announced its participation in PACK EXPO International, scheduled from October 18 to 21, 2026, where it will demonstrate its ColorWorks® on-demand color label printers, collaborative robots, and vision technology. These innovations aim to enhance the efficiency of packaging and fulfillment operations. The introduction of ColorWorks printers allows manufacturers to produce high-quality, customized color labels on demand, addressing the growing need for flexibility in labeling processes. Coupled with robotic automation, these solutions are designed to streamline various tasks, including handling, inspection, and packaging, thereby improving overall productivity. As the event approaches, industry professionals should watch for how Epson's integrated solutions can transform labeling and packaging workflows. No further timeline was disclosed at the time of publication.
OpenAI's GPT-6 Astra Dominates SuperCLUE's Robot-Brain Benchmark with Planning Superiority
OpenAI's GPT-6 Astra achieved the highest overall score in SuperCLUE's September 2026 EmbodiedCLUE-VLA evaluation, leading with a score of 85.71 in interaction and planning. This score significantly surpasses the runner-up, Gemini-3.8-Flash, which scored 64.29, highlighting Astra's notable advantage in planning capabilities. The importance of this evaluation lies in its comprehensive assessment of ten models, which includes basic perception, visual reasoning, interaction, planning, and embodied safety. SuperCLUE's report indicates that while Astra leads in overall performance, the evaluation's fixed input/output protocol may limit the generalization of results across different tasks and protocols, making it a useful but narrow comparison of each model's capabilities. Looking ahead, the next critical aspect to observe is how these models perform in real-world scenarios, particularly in executing plans and adapting to unforeseen circumstances. The results from physical trials will be essential to determine the practical applications of these models in robotics, beyond the theoretical scores presented in this benchmark.
OpenAI China
Robotics and Physical AI Addressing Key Physical Security Challenges in 2026
In 2026, over $23 billion has been invested in enhancing the integration of robotics and artificial intelligence for physical security. This investment aims to address the increasing complexity of threats against critical infrastructure, such as data centers and manufacturing plants. The adaptive capabilities of these technologies are essential for organizations to protect their assets effectively. The intersection of cybersecurity and physical protection presents new challenges, as demonstrated by the Jaguar Land Rover incident, where cyber breaches led to significant operational disruptions. Organizations must adapt to a rapidly changing security landscape, where legacy systems can exacerbate vulnerabilities. The need for proactive security measures is more pressing than ever, particularly in critical industries. At GTC Taipei 2025, companies like Foxlink showcased innovations such as the “robot dog,” designed to enhance physical security through adaptive functions and smart sensors. This technology represents a shift towards a more proactive security environment, enabling organizations to respond to complex threats effectively. No further timeline was disclosed at the time of publication.
Artificial Intelligence Artificial Intelligence / Cognition Automotive Autonomous Mobile Robots (AMRs) Cameras / Imaging / Vision Manufacturing
Safety Considerations Will Shape the Future of Humanoid Robots
In a recent discussion, Rick Balzano, Vice President of Go-to-Market for AI Services, Safety and Robotics at TaskUs, highlighted the critical work required to make humanoid robots reliable. He emphasized that unlike large language models, which operate on digital data, humanoid robots must navigate and interact with the physical world, necessitating extensive real-world data collection and validation. Balzano pointed out that while humanoid robots are often marketed as versatile workers, they currently remain highly specialized machines disguised in human-like forms. He noted that significant advancements have been made in areas such as locomotion and perception, but the gap between demonstration and real-world application is substantial, often involving controlled environments and remote assistance. Looking ahead, the industry must focus on bridging this gap to enhance the reliability of humanoid robots. Balzano's insights underscore the importance of safety and data integrity in developing robots that can operate effectively in diverse environments. No further timeline was disclosed at the time of publication.
AI and Robotics Innovation
RobotToday Weekly September 21 – 25, 2026
Cognex buys RealSense, Qualcomm acquires PickNik, the IFR counts 5 million factory robots, AGIBOT deploys 300 robots at Chimelong, and Tekever raises $580 million.
China’s Dexterous Hand Industry, 2026 Part 1 — The Demand Picture
MIR Databank puts China dexterous hand shipments above 30,000 units in 2026, yet only 3.5% of 2025 demand came from factory production. Part 1 of 3.
China’s Dexterous Hand Industry, 2026 Part 2 — The Technology
Linkage, tendon, direct drive or hybrid — and why hardware is converging while the DL1–DL5 manipulation scale decides who wins. Part 2 of 3.
ROBOTTODAY WEEKLY BRIEFING August 24 – 28, 2026
This week in robotics: XPeng's Dogotix raises over $900 million at a $6.3 billion valuation, Unitree's post-IPO slump revives bubble talk, Beijing's World Humanoid Robot Games close with a shift to real-world tasks, and Teradyne sues JAKA at Europe's Unified Patent Court. August 24–28, 2026.
WRC 2026 Core Components: 137 Robot Parts Suppliers
The third report in the WRC 2026 series. Report 1 mapped all 248 exhibitors by category and Report 2 covered the 139 that build finished robots. This one covers the 137 that build what goes inside them.
WRC 2026 Exhibitor Map: 248 Robotics Companies Mapped
We classified 248 WRC 2026 exhibitors with the RobotToday taxonomy. Motion & Actuation leads at 75 companies, Humanoid at 72, and compute stays thin at 12. (Modified 2026-08-27)
WAIC Review 4/4: From Port AGVs to Hospital Night Shifts: What WAIC 2026 Reveals About Real Robot Deployments
From 530+ AGVs at a Singapore port to robots on Apple's line, WAIC 2026 speakers shared concrete data on where embodied AI robots are actually working today.
XINMING Unveils R216g 3D Vision AI Module at Shanghai Conference
At the 2026 Zhangjiang Embodied Intelligence Developer Conference, XINMING unveiled its latest innovation, the R216g 3D Vision AI module. This cutting-edge technology is specifically designed for humanoid robots and collaborative arms, boasting advanced spatial perception capabilities that ensure high precision in complex environments. The introduction of the R216g aims to significantly improve the efficiency and reliability of physical AI applications across a range of industries, reflecting XINMING's commitment to advancing robotics and artificial intelligence.
3D Vision AI Humanoid Robots Collaborative Robotics Spatial Intelligence Industrial Automation
