Meet Unitree Super Athlete AS2-W; China's New All-Terrain Wheeled Robot With Advanced Mobility (Watch Video)

latestly.com Jul 25, 2026

Meet Unitree Super Athlete AS2-W; China’s New All-Terrain Wheeled Robot With Advanced Mobility (Watch Video)

Curated by Shubham · TechCapsules · Jul 25, 2026

AI-summarised brief · reviewed before publication

Chinese robotics firm Unitree unveiled the Super Athlete AS2‑W, an all‑terrain wheeled robot capable of sprinting up to 6 m/s while maintaining stability after high‑platform drops. Weighing about 25 kg, the machine combines wheel speed with leg articulation to climb rocky slopes, uneven steps and shallow streams, thanks to an IP54 water‑resistance rating and a robust load‑carrying capacity for heavy payloads. Its navigation relies on a pre‑trained reinforcement‑learning model that enables autonomous movement through complex environments. The demonstration, released on July 25, 2026, highlighted the robot’s ability to traverse rugged landscapes and drop onto platforms without loss of control. Observers note the platform’s potential for both civilian field work and defence applications, marking a notable leap in China’s unmanned‑system capabilities.

💡 Why It Matters

· The AS2‑W showcases a new speed‑and‑agility benchmark that could redefine how autonomous robots operate in harsh, real‑world settings.

Capsules Robotics

Read full article on latestly.com

Optical Tech Would Update a Robot’s AI on the Fly

Cornell Tech researchers Yifan He and Jae-sun Seo presented a novel optical receiver design at the IEEE/JSAP Symposium on VLSI Technology & Circuits that updates AI model parameters directly via light. The system uses digital QR-code-like matrices to flip binary values in static random-access memory cells containing photodiodes, bypassing power-hungry analog circuits. This approach addresses the energy inefficiency and cost bottlenecks of moving data between dynamic random-access memory and processors using electrical connections. By enabling fully digital optical communication, the technology aims to reduce energy consumption in data centers, self-driving cars, and edge applications like robots. The current lab prototype uses a static light mask, but the team plans to develop transmitters capable of rapid matrix changes. University of Michigan’s Dennis Sylvester praised the solution’s cleverness and commercial potential, though he noted that larger photosensitive bit cells may currently limit memory density compared to conventional chips.

💡 The technology tackles the memory wall by eliminating energy-intensive data transfers between storage and processing units. This direct optical writing could significantly lower power demands for scaling AI workloads in energy-constrained environments.

en.sedaily.com · Jul 26

Incisionless Korean Surgical Robot Clears FDA, Signs Exclusive Olympus Deal

Endorobotics, a Korean medical robotics firm, has secured U.S. Food and Drug Administration approval for its ROBOPERA surgical robot and signed a global exclusive distribution agreement with Olympus. ROBOPERA distinguishes itself by attaching robotic arms to existing diagnostic endoscopes, enabling incisionless procedures through the mouth or anus. This approach eliminates the need for large-scale surgical equipment, reducing hospital adoption costs. The device enhances precision in complex endoscopic submucosal dissection for early-stage cancers by stabilizing tissue manipulation. Endorobotics became the first Korean company to receive the "NAY" product code for robotic control surgical systems, matching the classification of Intuitive Surgical’s da Vinci robot. The firm previously raised 32.5 billion won in Series C funding and was recognized by the Korea Innovation Foundation for successful commercialization of public research achievements.

💡 By leveraging existing endoscopy infrastructure, this technology drastically lowers the financial barrier for hospitals to adopt robotic surgery. It challenges the dominance of traditional incision-based systems by offering a more accessible, precision-driven alternative for gastrointestinal treatments.

therobotreport.com · Jul 26

Magmotor makes motors for a changing world for 150+ years

Magmotor Technologies Inc., founded in 1876 as Kendrick and Davis, celebrates its 150th year while announcing a expansion that aims to triple revenue and staff. The Worcester, Massachusetts firm continues to produce brushed AC and DC fan motors for aerospace, defense and high‑tech customers, many of which have operated for 15‑20 years without failure. Long‑tenured employees such as production manager Jamie Willigar and design engineer Scott Cerra underscore the company’s deep expertise. Although most manufacturing shifted to Chinese partners after the 1990s, all design and final assembly remain in‑house, and Magmotor plans to re‑introduce limited U.S. production. Its brush motors power NASA rovers, BattleBots, ship‑cleaning robots and other niche applications, and the firm is exploring new AC and brushless models.

💡 By scaling up while reshoring key assembly, Magmotor positions itself to meet rising demand for reliable brushed motors in emerging robotics and space missions, securing a niche that many competitors have abandoned.

en.sedaily.com · Jul 26

KAIST, Nvidia Deepen AI Alliance With New Physical AI Center

KAIST and Nvidia announced the establishment of the Human Physical Intelligence Technology Center to develop physical artificial intelligence capable of understanding human movement and intent. The center, co-directed by Professors Kong Kyoung-chul and Kim Jung, aims to build a Human Motion Foundation Model. This model will predict human actions by analyzing large-scale data on walking, joint movements, and balance. Unlike pre-programmed robots, those using this technology will adjust to user posture in real time, benefiting wearable robots, humanoids, and digital healthcare. KAIST contributes its extensive real-world motion data and motor control expertise, while Nvidia provides technical support via its Omniverse platform. This collaboration complements a separate joint institute focused on agentic AI for Korean industries. The initiative addresses the unique data requirements of physical AI, which relies on real-environment movement data rather than internet-based text or images.

💡 Physical AI requires real-world motion data, a scarce asset that KAIST possesses, giving this partnership a distinct competitive edge over text-based AI models. By enabling robots to predict and adapt to human movement, the technology bridges the gap between digital intelligence and physical interaction.

autonews.gasgoo.com · Jul 25

Embodied Intelligence Data, How Long Until Supply Is Sufficient?

The embodied intelligence sector showcased a dramatic expansion at this year’s WAIC in Shanghai, with exhibitor numbers rising from 18 in 2022 to over 200 in 2026. Despite the proliferation of data‑collection tools—teleoperation rigs, wearable sensors, body‑free setups and high‑fidelity simulators—the industry faces a severe data shortage. Global high‑quality embodied interaction recordings total only about 500,000 hours, far below the roughly 10 million hours required to train general‑purpose embodied models, a gap of more than 99 percent. Companies such as Agile Robots, ROBOTERA, Encos Intelligent Technology, GenRobot and OYMotion demonstrated diverse approaches, from millisecond‑level force‑feedback teleoperation to 1 kHz glove sensors and full‑body motion capture, yet the fundamental supply‑demand imbalance remains unresolved.

💡 Without closing the data gap, the promise of universally adaptable robots will stall, keeping advanced automation confined to niche, heavily curated applications.

therobotreport.com · Jul 25

How time series databases unlock real-time data for robotics

Episode 254 of The Robot Report Podcast features Doug Pagnutti, developer advocate at Tiger Data, discussing how time series databases like TimescaleDB enhance industrial automation, robotics, and AI applications. Pagnutti, who bridges the operational technology and information technology gap, shares insights on integrating these databases with various sensors, managing data at scale, and optimizing performance across cloud and edge environments. With over twelve years of experience in oil and gas and manufacturing automation, including work on PLCs, robots, and SCADA systems, Pagnutti helps automation engineers tackle time-series data infrastructure for large-scale industrial internet of things deployments. The episode also covers recent industry news, including China’s Unitree stating that a generative pre-trained transformer moment for robots remains years away, Vicarious Surgical officially shutting down, and U.K.-based Humanoid securing $152 million in Series A funding. Tiger Data, the episode sponsor, promotes TimescaleDB as an extension of Postgres designed to handle live data workloads without requiring additional pipelines or migrations.

💡 As robotics systems generate massive volumes of sensor data, specialized infrastructure becomes critical for maintaining real-time responsiveness without overwhelming standard database architectures. This technical shift enables manufacturers to deploy scalable, low-latency solutions that directly support the operational demands of modern industrial automation.