China Page 5

Inclusion 2026, officially the 2026 Inclusion · Conference on the Bund, runs Sept. 9–12 in Shanghai under the theme “Building the AI Economy Together”. Hosted annually by Ant Group since 2020 with official government support, the four-day event convenes over 600 global academics, entrepreneurs and investors, alongside 300 tech exhibitors, cementing its international standing as a key AI industry summit. The event comprises a 15,000-square-metre tech exhibition, one main forum and 40 themed sessions, plus Developer Day, VC meetups and a new tech talent fair. It also debuts its first AI Art Festival, merging tech innovation with digital art creation. Top global speakers headline the conference, including Nobel laureates Thomas Sargent and Philippe Aghion, leading Chinese academicians, and executives from...

Chinese electric vehicle maker Xpeng Inc. said on Tuesday it has commissioned its humanoid robot production line, with the first Iron unit completing automated assembly and walking off the line autonomously as the company transitions from prototype development to manufacturing. The line, which the company claims is the world’s first automated production line for advanced humanoid robots, achieves over 80% automation in core processes and integrates automotive-grade quality control systems to support future mass production and capacity expansion. The Iron humanoid features 76 degrees of freedom across its body and 21 degrees of freedom in each hand, with a fully enclosed flexible lattice structure designed for both human-like appearance and safety. Powered by three in-house Turing AI chips delivering 2,250...

Maniformer, a Shanghai-based physical AI data platform, won the All-Scenario Data Collection Technology Innovation Award at the Global Product Technology Innovation Awards during the IFA 2026 consumer electronics fair here, marking the first time a Chinese physical AI data infrastructure company has received recognition at the century-old exhibition. The awards, established in 2014 by International Data Group (IDG) and the German Chamber of Commerce and Industry (DIHK) with annual support from IFA, recognise product technology innovation and industrial application value. Other winners this year include Samsung, Sony, LG, Siemens and Dyson. Maniformer addresses what the industry sees as a critical bottleneck: as robots accelerate from labs to factories, commercial spaces and homes, scaling embodied AI models requires massive, high-quality real-world...

ACE ROBOTICS, in collaboration with the Chinese University of Hong Kong’s Multimedia Laboratory and Nanyang Technological University, has open-sourced ACE-Ego-Hand, a 3D dual-hand shape reconstruction model designed for first-person video, generating approximately 5,000 hours of training data that could serve as a foundational dataset for embodied AI model training. The model extracts structured 3D bimanual manipulation trajectories from massive egocentric video footage, translating human grasping, assembly and hand-coordination actions into machine-learnable data. It marks the first time a video generation model has been repurposed as a DreamHand encoder for end-to-end training, outputting hand pose, shape and 3D position in a single forward pass without frame-by-frame detection or temporal regression. Efficiency gains over traditional approaches Unlike previous hand models reliant on...

Tingbo He, head of semiconductor research at Huawei Technologies, published a new academic paper on Thursday addressing concerns about heat dissipation in 3D-stacked chips, marking the company’s third paper on its “τ-law” chip design philosophy in less than four months. The latest paper, titled “Huawei’s τ Chip Was Supposed to Melt?”, responds to scepticism that folding circuits into three dimensions would inevitably lead to higher power density and thermal issues. “Heat is the most troublesome hidden concern of the τ-law,” She wrote. “Inside the chip, the main energy consumption does not come from computation itself, but from data transfer.” Contrary to expectations that circuit folding would dramatically increase transistor density and power density, the paper argues that folding actually reduces...

Xiaomi’s CyberOne humanoid robot drew crowds of visitors and media at its first overseas exhibition during the IFA consumer electronics fair here, with the robot responding to onlookers with waves and victory signs. The Chinese tech giant originally unveiled the full-size humanoid bionic robot, internally codenamed “Tie Da”, in August 2022. The first-generation model stands 177 cm tall, weighs 52 kg, and features 21 degrees of freedom with 0.5-millisecond real-time response, powered by Xiaomi’s self-developed bionic perception, biomechatronic fusion and artificial intelligence technologies. This year, Xiaomi introduced a new-generation humanoid robot, also named “Tie Da”, which made its first public appearance at the World Robot Conference 2026 in Beijing. The new version stands approximately 170 cm tall, weighs 66 kg,...

AGIBOT, a pioneer in the global general-purpose AI robotics industry, is showcasing its latest robot portfolio at IFA 2026, led by its third-generation full-size humanoid, the AGIBOT A3, with applications spanning retail, industrial operations and commercial cleaning. The showcase also features the AGIBOT X2 Ultra on IFA’s “Robots on the Runway,” a new TÜV Rheinland certification for the AGIBOT A-Series, and a joint showcase with European partner Tekpoint as AGIBOT expands its commercial footprint in Europe. AGIBOT A3 Leads a Multi-Robot Showcase At the center of AGIBOT’s showcase is the A3, a 173-centimeter, 55-kilogram full-size humanoid integrating locomotion, interaction and manipulation intelligence. The robot combines dynamic mobility, multimodal interaction, hot-swappable batteries and coordinated multi-robot operation. The AGIBOT X2 Ultra will...

PrimeBOT, a consumer embodied AI brand under SWANCOR, made its IFA debut on Thursday with two “personal robots” aimed at bringing robotics into households, a move the company said prioritises affordability and user engagement over industrial-grade performance. The T1 and Q1 models are designed to be compact, lightweight, and priced to lower the barrier for home adoption, according to the company. At its booth, European parents were observed allowing children to interact freely with the robots, reflecting confidence in their safety and approachable design. Beyond emotional companionship, the robots are positioned for practical daily uses including camping tracking, towing carts, and load-carrying, the company said. Modular design and co-creation draw crowds PrimeBOT’s booth, themed “PrimeGallery,” features a multi-language photo wall...

Chinese robotics firms LEJU Robot and Heyman have jointly launched a scene-oriented solution centred on a general-purpose humanoid robot body and a multi-fingered dexterous hand, aiming to eliminate the need for custom end-effector swaps across diverse industrial tasks. The system addresses common pain points such as trade-offs between payload and precision, low gesture-development efficiency, and difficult redeployment, the companies said. The integrated dexterous hand features a variable-configuration design that supports seven plus grip modes – including two-finger pinch, internal expansion, enveloping grasp, and two-handed hold – on the same hardware. It delivers a payload of at least 24 kg and a repeated positioning accuracy of ≤0.03 mm, outperforming the industry norm of 0.2 mm. This allows a single manipulator to...

Astribot, a Chinese robotics startup, has unveiled SmoothRL, an online reinforcement learning framework designed to train policies during asynchronous execution – a growing reality for embodied AI deployments where inference and action run at different frequencies. The work, led by the company’s foundation model team, addresses a fundamental mismatch: when a robot receives new action chunks before previous ones finish, conventional RL that updates on entire chunks ends up crediting or blaming actions that were never actually executed. The framework partitions each action chunk into three regions – committed (already fixed), execution (actually performed), and discarded (overwritten by later inference) – and backpropagates gradients only through the execution region. This ensures that optimisation mirrors the robot’s real experience. SmoothRL also...