Middle East - OUT2ALL - Recent Post in Robotic

Unitree UnifoLM‑X2‑1.0 Powers Humanoid Robot for World‑First Fully‑Autonomous Combat Demonstration

Mon, 07 Sep 2026 16:03:43 +0000

7 September 2026, Unitree Robotics (688836) released demonstration footage showcasing the world‑first fully‑autonomous humanoid robot combat driven in real‑time by its in‑house World‑Action foundation model, UnifoLM‑X2‑1.0. Without human remote control or pre‑scripted fighting routines, the humanoid robot runs a full closed‑loop workflow of perception, physical‑world prediction, offensive‑defensive decision‑making and motion execution entirely powered by AI.
Previous robot fighting displays mostly relied on pre‑programmed motion libraries or human operators behind the scene. In this new showcase, the robot responds dynamically to random strikes and movements from its opponent. It detects postures, anticipates incoming moves, and independently generates punches, blocks, dodges and footwork. After physical impacts, it actively adjusts posture and regains balance. All responses are computed on‑the‑fly rather than triggering pre‑recorded motion clips.
UnifoLM‑X2‑1.0 is part of Unitree’s UnifoLM World‑Action model family. It addresses key bottlenecks for high‑dynamic physical interaction: instant planning, real‑time decision‑making and responsive execution. The model carries internal physical simulation. Instead of only interpreting visual input, it forecasts upcoming states of opponents and physical collisions, and formulates tactics in advance. It builds a complete embodied‑AI pipeline: perception‑world prediction‑motion planning‑whole‑body actuation.
Combat is a high‑stress test case featuring fast rhythms, frequent impacts and constant balance disturbances. Unitree clarifies that fighting itself is only a technical validation scenario, not a commercial product direction. Insights gained will translate to household and industrial deployments, enabling robots to cope with unforeseen physical disturbances in real‑world settings and advance large‑scale practical adoption of world‑model‑driven humanoids.
While marking a leap from scripted performance to genuine autonomous interaction, the demo remains a lab‑stage achievement. Full technical papers and independent third‑party evaluation are yet to be published, and substantial gaps remain before mass commercial deployment.
(End)

SpaceX‑Tesla Space‑Ground Integrated Remote‑Sensing Program Gets Physical as Cybercab Makes Its Debut, Building the World’s First Vehicle‑Mounted Mobile Space‑Awareness Network

Thu, 03 Sep 2026 19:02:20 +0000

Release Date: September 4, 2026Source: Commerce Wire Global Technology Industry ChannelIndustry Sectors: Space‑Tech, AI Autonomous Driving, Spatial Remote Sensing, Digital Mapping InfrastructureHighlightsWith mass‑produced Cybercab vehicles fully equipped with factory‑integrated Starlink V5 on‑board satellite terminals, the long‑anticipated SpaceX‑Tesla space‑ground integrated remote‑sensing program welcomes its sole ground‑based operational entity. Breaking away from static remote‑sensing models relying on top‑down satellite observation, Cybercab deploys a fleet of millions of autonomous vehicles as mobile ground‑based remote‑sensing nodes, closing the loop of “space‑satellite relay — on‑vehicle AI perception — cloud‑based data interpretation”. This marks the official arrival of a new era of dual‑space dynamic surveying within the global spatial remote‑sensing industry.1. Industry Milestone: The Remote‑Sensing Program Moves from Concept to Physical HardwareConventional commercial remote‑sensing systems have long depended on aerial surveys and scheduled imagery captured by low‑Earth‑orbit satellites, which suffer from slow update cycles, insufficient ground‑level granularity and limited capacity for complex road‑condition perception. High‑precision maps worldwide are generally updated on a monthly or quarterly basis, failing to meet real‑time demands for autonomous driving and smart‑city development.The space‑ground integrated remote‑sensing program built through strategic collaboration between SpaceX and Tesla reshapes traditional remote‑sensing architecture:• Space‑based Layer: Comprises SpaceX’s third‑generation Starlink constellation, the Starshield commercial remote‑sensing system and the upcoming Starmind AI remote‑sensing satellites, tasked with large‑scale Earth observation and universal data relay transmission.• Ground Operational Layer: The Cybercab robotaxi fleet serves as the exclusive dedicated execution entity. Unlike passive data collection from ordinary consumer Tesla vehicles, Cybercab enables controlled, continuous and high‑density urban mobile remote‑sensing operations.In August 2026, real‑world vehicles at the Texas Gigafactory were publicly revealed, featuring concealed roof‑mounted Starlink V5 terminals factory‑fitted instead of retrofitted. This development confirms the finalisation of the program’s hardware architecture and the official launch of its commercial‑grade operational framework.2. Core Positioning: Cybercab as the Dedicated Execution Platform for the Space‑Ground Remote‑Sensing ProgramIndustry analysts clarify that consumer‑facing Model‑series vehicles only act as supplementary data‑collection endpoints, while Cybercab functions as the primary operational entity of the space‑ground integrated remote‑sensing initiative. Three irreplaceable core attributes support its professional‑grade remote‑sensing operations:2.1 Globally Controllable Fleet‑Dispatching SystemCybercab forms Tesla’s in‑house robotaxi commercial fleet managed via a unified backend with round‑the‑clock shift‑based operations. Distinct from random travel patterns of privately owned vehicles, the platform may actively dispatch vehicles to designated zones for targeted remote‑sensing surveys in response to road renovations, infrastructure upgrades and new‑district development, enabling precise completion of urban spatial‑data coverage.2.2 Mass‑Produced Space‑Communication Hardware as Standard EquipmentProduction‑specification Cybercab units come factory‑installed with lightweight Starlink V5 satellite terminals weighing merely 1.1 kg. Compared with prior‑generation hardware, volume is reduced by 48 % and power consumption drops to 35–50 W, striking a balance between seamless roof‑body integration and operational stability under mobile conditions. The hardware delivers a peak download speed of 375 Mbps, surpassing mainstream residential broadband connections. It efficiently handles compressed remote‑sensing image transmission, fleet telemetry and high‑frequency map over‑the‑air updates, fundamentally addressing the challenge of data upload from suburban, mountainous and remote‑town cellular‑network dead zones.To facilitate stable satellite‑signal transmission, Tesla has optimised roof‑panel polymer structures to eliminate electromagnetic shielding caused by metallic vehicle bodies and sustain reliable space‑to‑ground link connectivity while vehicles are in motion.2.3 AI Edge Computing Empowers Intelligent Remote‑SensingThe on‑vehicle AI4 supercomputer runs a full‑scene visual‑perception framework. Leveraging an eight‑camera FSD high‑definition vision array, it captures fine‑grained spatial data including road markings, traffic installations, building outlines and pavement conditions in real‑time.Unlike full‑volume data transmission adopted in traditional remote‑sensing workflows, Cybercab carries out pre‑screening via on‑vehicle AI. It instantly compares live footage against historical high‑definition map datasets, identifies key frames capturing environmental changes and transmits compressed outputs only. This drastically conserves precious satellite bandwidth and establishes an efficient remote‑sensing mechanism defined by “edge‑end screening — on‑demand transmission — cloud‑based updates”.3. Complete Operational Logic: End‑to‑End Space‑Ground Remote‑Sensing WorkflowFollowing Cybercab’s commercial roll‑out as an operational asset, the SpaceX‑Tesla remote‑sensing program implements standardised closed‑loop workflows to deliver round‑the‑clock automated spatial‑data updates:1. Parallel Survey Operations: While performing urban passenger‑transport duties, Cybercab automatically conducts mobile road‑level remote‑sensing data collection covering primary and secondary urban arteries, emerging districts and township road networks.2. AI‑Driven Scene‑Change Detection: The on‑board computer analyses real‑time scene variations, filters redundant footage and pinpoints map‑modification signals such as road resurfacing, facility additions and route adjustments.3. Dual‑Mode Intelligent Transmission: Data is uploaded preferentially over 5G/LTE cellular networks. Automatic sub‑second switching to the Starlink V5 satellite link preserves data continuity when vehicles enter communications‑blackout areas.4. Dynamic Cloud‑Based Map Refresh: Remote‑sensing datasets feed into Tesla’s global mapping platform to enable high‑definition‑map refreshes at a daily frequency — far outpacing the monthly update cycles of traditional mapping vendors.5. Future‑Oriented Space‑Ground Data Fusion: Once SpaceX launches the first batch of Starmind space‑based AI remote‑sensing satellites in Q4 2027, fine‑grained ground observations collected by Cybercab will be fused with satellite‑captured multi‑spectral macro‑imagery. xAI will perform comprehensive data interpretation to deliver fully‑integrated space‑ground remote‑sensing capabilities.4. Technical Clarification: Satellite Links Not Used for Autonomous‑Driving Safety‑Critical ControlsTesla’s engineering leadership has formally clarified functional boundaries: Starlink satellite communications do not support safety‑critical decision‑making for autonomous driving.Perception, judgement, route planning and vehicle control for Cybercab’s Level‑4 autonomous operations run entirely on local on‑board hardware and edge‑AI processing, independent of satellite‑network connectivity. Within the remote‑sensing ecosystem, Starlink exclusively serves as a backup data‑transmission channel dedicated to remote‑sensing data uploads, remote fleet monitoring and global map upgrades. This strict separation between mobility‑safety and data‑service workflows safeguards the operational stability of self‑driving vehicles.5. Industry Disruption: Reshaping the Global Remote‑Sensing and Smart‑Mobility LandscapeCybercab’s launch as the operational backbone of the remote‑sensing initiative breaks long‑standing technological monopolies in the traditional remote‑sensing sector and delivers three layers of transformative industry value:5.1 Building an Exclusive Moat for Autonomous‑Driving TechnologyCurrently, all global autonomous‑driving manufacturers rely on third‑party commercial map data, characterised by update lags and limited scenario adaptability. By combining a satellite constellation with a fleet of millions of mobile remote‑sensing vehicles, Tesla builds a self‑governed dynamic‑mapping ecosystem optimised for complex urban road conditions and remote geographies, laying critical groundwork for widespread Level‑4 autonomous‑driving deployment.5.2 Opening a New Track for Commercial Remote‑SensingLegacy commercial remote‑sensing predominantly serves land‑surveying, resource‑exploration and disaster‑monitoring use‑cases with narrow application scope. The SpaceX‑Tesla space‑ground remote‑sensing ecosystem can deliver granular datasets tracking urban‑infrastructure dynamics, road‑network evolution and shifts in built‑up‑area spatial patterns, unlocking new commercial‑monetisation models for B2B markets spanning smart‑city governance, transport planning, logistics infrastructure and mapping services.5.3 Establishing Future Standards for Spatial‑Data AcquisitionThe hybrid framework combining vehicle‑borne mobile remote‑sensing and low‑Earth‑orbit satellite macro‑observation addresses the long‑standing industry gap described as “satellites capture the macro‑view, ground‑based platforms lack fine‑level detail”. It sets new technical benchmarks for global smart spatial‑data collection, updating and interpretation workflows.6. Current Challenges and Industry ConstraintsNotwithstanding the official roll‑out of the integrated system, the space‑ground remote‑sensing program faces several phase‑specific limitations:1. Satellite‑Bandwidth Cost Constraints: High transmission costs for low‑Earth‑orbit satellite mobile‑services currently support transfer of AI‑filtered critical datasets only. Large‑scale commercial adoption requires further optimisation of cost‑engineering models.2. Cross‑Border Data‑Regulatory Barriers: On‑vehicle street‑level remote‑sensing data capture and cross‑border satellite‑mediated transmission must comply with national surveying‑and‑mapping, data‑security and privacy‑protection legislation. Global roll‑out demands jurisdiction‑by‑jurisdiction compliance approvals.3. Fleet‑Production Ramp‑Up Pressures: Built around the innovative Unboxed manufacturing process, Cybercab’s production‑scale expansion directly determines coverage density and operational capacity of the ground‑based remote‑sensing network.7. Concluding Industry CommentaryWith Cybercab’s official debut as the dedicated operational entity, the SpaceX‑Tesla space‑ground integrated remote‑sensing program has definitively moved beyond conceptual‑stage development to become the world’s sole dual‑tier perception system delivering “macro‑scale space‑based remote‑sensing plus fine‑resolution mobile ground‑level surveys”.This technology‑driven collaboration not only rewrites the underlying competitive logic governing high‑definition mapping and autonomous‑driving industries, but also ushers in an era in which consumer‑grade mobility assets participate in professional spatial remote‑sensing. Further deployment of the Starmind satellite constellation and accelerated Cybercab mass production will soon bring a globally‑covering, real‑time‑updating, space‑ground‑fused next‑generation intelligent spatial‑awareness network into full operation.Commerce Wire Technology Industry Deep‑Dive Research Centre

Hugging Face Launches Microduck Open‑Source Biped Robot at USD399, Pre‑Orders Hit USD2.6M Within First‑Day of Launch

Wed, 02 Sep 2026 23:00:02 +0000

PARIS, Aug.27, 2026 /Commerce Wire News/ — Hugging Face, world‑leading open‑source AI platform, has officially launched Microduck, an open‑source miniature biped robot, through its French robotics subsidiary Pollen Robotics. Priced at USD399 for global developer pre‑orders, the new product has generated overwhelming market traction, securing USD2.6 million in confirmed pre‑orders within 24 hours. Peak sales hit one unit every four seconds. Official delivery guidance has since been revised to a four‑to‑six‑month lead‑time for new orders amid supply‑chain constraints.
Standing 25 cm tall and weighing under 800 g, the duck‑shaped Microduck integrates 15 high‑precision servo motors, wide‑angle camera, ToF LiDAR and dual IMUs. Powered by Rockchip RK3566 edge‑AI system‑on‑chip with built‑in NPU, it supports on‑device AI inference. Connectivity includes Wi‑Fi, Bluetooth and NFC. Its duck‑bill gripper delivers a payload capacity up to 800 grams. Seven preset behaviours such as walking, squatting, object‑grasping and fall self‑righting are available out‑of‑the‑box.
The platform delivers an end‑to‑end reinforcement‑learning workflow for robot developers. Programmers can train locomotion‑control policies in the MuJoCo physics simulation environment via Python APIs, export trained models in ONNX format for physical‑robot deployment, and publish finished projects onto Hugging Face Hub, enabling a full pipeline of simulation, training, deployment and community‑driven knowledge sharing.
Clem Delangue, Chief Executive Officer of Hugging Face, remarked at the product launch that Microduck is designed to become the “Raspberry Pi of embodied intelligence”. Historically, high‑priced biped‑robot research hardware has created barriers for universities, independent developers and start‑ups conducting experiments on world‑models and multimodal human‑robot interaction. By combining an affordable price tag with fully‑open architecture, Microduck lowers barriers to physical‑AI R&D and fosters distributed innovation worldwide.
Hugging Face completed its acquisition of Pollen Robotics back in April 2025 to enter the developer‑robotics market, following the Reachy Mini desktop robot product series. Unlike consumer‑grade household robots, Microduck is built purely as an open R&D and algorithm‑validation platform, targeting university labs, AI research groups, freelance engineers and robotics start‑ups.
Sales metrics indicate pre‑order revenue passed USD1 million six hours post‑launch, climbing to USD2.6 million after 24 hours. Pollen Robotics published an official supply‑chain update on its website: new customers can no longer guarantee Christmas‑2026 delivery. Orders are fulfilled on a first‑come‑first‑served basis, with revised delivery lead‑times of four‑to‑six months. By early September, cumulative global reservations have surpassed 10 000 units, drawing widespread attention from the international tech sector, open‑source community and venture‑capital investors.
Industry analysts note that Microduck’s commercial breakthrough marks a major inflection‑point for the embodied‑AI industry. Humanoid‑robot research has long been dominated by large technology corporations with substantial capital expenditure requirements. Microduck brings together European robotics algorithms, Chinese semiconductor solutions and mature Asian precision‑manufacturing supply‑chains, delivering a standardized, low‑cost hardware foundation. This empowers large cohorts of developers to conduct research on locomotion control, on‑device world‑modelling and sim‑to‑real transfer, accelerating AI’s transition from large‑language‑models toward real‑world physical deployment.
As the open‑source biped‑robot ecosystem matures, a dual‑track industry roadmap is expected to take shape in coming years: large‑scale enterprises will focus on commercial‑grade general‑purpose humanoid robots, while smaller developer teams will leverage platforms such as Microduck to run niche‑use‑case algorithm trials. These parallel innovation pathways are poised to jointly drive the expansion of the global embodied‑AI sector.
Key Specifications
• Developer: Pollen Robotics (Hugging Face Group)
• Official Price: USD 399
• Height: 25 cm
• Weight: <800 g
• Actuators: 15 high‑precision servo motors
• Sensors: Wide‑angle camera, ToF LiDAR, dual IMUs
• Main Chip: Rockchip RK3566 with built‑in NPU for edge‑AI inference
• Software Stack: Python API, MuJoCo simulation, ONNX deployment, Hugging Face Hub
• Delivery Update: Original Christmas‑2026 batch closed; new‑order lead‑time: four‑to‑six months
About Hugging Face
Headquartered in the United States, Hugging Face is one of the world’s largest open‑source‑AI community platforms, hosting large‑language‑models, multimodal models, robotics development toolkits and other resources. Its LeRobot initiative is dedicated to advancing open‑source innovation across robotics.

SoftBank in Talks to Acquire Majority Stake in Humanoid Robot Firm 1X Technologies at $6 Billion Valuation

Sat, 29 Aug 2026 08:43:40 +0000

According to industry sources cited by The Information on August 27, SoftBank Group is in exclusive negotiations to acquire a majority stake in Norway-based humanoid robot startup 1X Technologies. The proposed deal values the company at approximately $6 billion.
Negotiations are still ongoing and deal terms remain subject to adjustment. No definitive agreement has been signed, and neither SoftBank nor 1X Technologies has officially commented on the market rumours.
Founded in 2014 and headquartered in Oslo, 1X Technologies — formerly known as Halodi Robotics — counts OpenAI Startup Fund and NVIDIA among its key institutional investors. The company focuses specifically on consumer and household humanoid robots, a differentiated positioning in the global humanoid industry.
Its flagship bipedal robot model NEO is designed for daily domestic scenarios, capable of folding clothes, cleaning residences, watering plants, and handling routine household tasks. Priced at $20,000 for full purchase, the robot is also available via a $499 monthly subscription plan. Pre-orders opened in October 2025, with official deliveries originally scheduled for 2026, though mass commercial shipment has not yet commenced.
The $6 billion valuation marks a notable shift in the company’s financing trajectory. In late 2025, 1X Technologies aimed to raise $1 billion at a $100 billion valuation but secured far less capital than targeted. While the current deal valuation represents a downward revision from its last fundraising ambition, it still reflects a substantial upgrade from its $820 million valuation recorded in early 2025. As the firm remains in the early commercialisation stage with limited revenue scale, SoftBank’s potential investment will provide substantial capital and computational resources to accelerate hardware iteration and AI model optimisation for humanoid applications.
The proposed acquisition also signals SoftBank’s official return to the core humanoid robot track. The move comes merely one month after SoftBank fully exited Boston Dynamics by selling its remaining stake, ending its previous bet on industrial-grade humanoid robots.
SoftBank’s strategic approach has undergone a comprehensive upgrade after its earlier setbacks with the Pepper robot and Boston Dynamics investment. The Group is now building a full-stack Physical AI industrial ecosystem. It previously acquired ABB’s robotics business for $53.75 billion to secure industrial hardware manufacturing capacity and led a $1.4 billion financing round for Skild AI to strengthen underlying robot foundation model capabilities.
A successful majority acquisition of 1X Technologies will complete SoftBank’s industrial layout, covering industrial robot production, general robotic AI algorithms, and consumer-facing home humanoid terminals, fully implementing Masayoshi Son’s long-term Physical AI vision.
Masayoshi Son, founder of SoftBank, previously projected that the global humanoid robot market will exceed HK$290 trillion by 2040, with over one billion AI-powered humanoid robots deployed across industrial, commercial and household scenarios worldwide. The investment in 1X Technologies represents SoftBank’s core bet on the mass adoption of consumer humanoid robots in ordinary households.
Market analysts point out that the global humanoid robot industry is currently at a critical inflection point. While most startups face pressure from delayed mass production, slow delivery progress and tightened capital financing, leading industrial players are making counter-cyclical mergers and acquisitions, accelerating industry resource concentration toward powerful technology and capital platforms.
If closed smoothly, the $6 billion transaction will become SoftBank’s largest majority investment in the humanoid robot sector to date, and is expected to significantly accelerate the global commercialisation process of household humanoid robots.
Disclaimer: This report is for informational purposes only and does not constitute any investment advice.

Unitree Robotics Unveils New Humanoid Robot “Superman”, Surpassing Human Physical Limits in Speed and Jumping Performance

Thu, 20 Aug 2026 13:15:09 +0000

Hangzhou, August 17, 2026 — Unitree Robotics has officially launched its brand-new high-performance humanoid robot, internally codenamed “Superman”, following an ultra-fast R&D cycle of just over three months. The new model delivers groundbreaking improvements in explosive power and locomotion speed, comprehensively exceeding the maximum physical limits of human running and standing high jump capabilities.
Developed and validated in merely three months, the “Superman” humanoid represents a major leap in high-dynamic bipedal robotics. Despite a compact body design with a leg length of only 0.85 meters, the robot achieves a standing high jump of 2 meters and a peak sprint speed of 12.66 meters per second. Both key performance indicators surpass the highest recorded athletic benchmarks of elite human athletes, marking a new milestone in high-speed, high-explosiveness humanoid robot development worldwide.
The outstanding motion performance is supported by Unitree’s self-developed high-power-density joint actuators and advanced reinforcement learning motion control algorithms. The technical breakthrough breaks the long-standing industry bottleneck that most humanoid robots can only achieve low-speed movement and limited jumping ability. It verifies that electric-driven bipedal robots can transcend human biological motion boundaries in extreme athletic scenarios.
Different from Unitree’s existing H-series and G-series general-purpose humanoid robots focused on daily operation and refined manipulation, the “Superman” is positioned as a professional high-dynamic motion research and verification platform. It is specially built for exploring extreme bipedal movement technologies, including high-speed continuous running, ultra-high jumping, and adaptive dynamic balance under intense motion.
Unitree Robotics stated that the current “Superman” prototype is still in the rapid iteration stage. The R&D team will continue to optimize its motion stability, battery endurance, terrain adaptability and environmental robustness in subsequent upgrades. The underlying motion control technologies verified by this project will be fully applied to Unitree’s full lineup of humanoid and quadruped robots, empowering industrial inspection, emergency rescue, special operation and high-end robotic application scenarios.
Industry analysts commented that the release of the “Superman” humanoid opens a new technical track for high-dynamic bipedal robots. It not only proves the extraordinary potential of electric-driven humanoid robots in extreme motion capabilities, but also provides valuable technical references for global embodied intelligence and bionic robotics research.
With the expansion of its high-performance humanoid product matrix, Unitree Robotics further consolidates its leading position in the global robotics industry. The company will continue to open core motion control technologies and carry out in-depth cooperation with global academic institutions and industrial partners to promote the continuous breakthrough and commercial implementation of high-end bipedal robot technologies.