10 results on this page
Tactile robotics is expanding from sensor hardware into datasets, foundation models, and evaluation benchmarks. Research shows touch improves physical reasoning and robot success, but inconsistent sensor formats create major cross-device transfer challenges.
Humanoid robot traffic officers in China can direct vehicles, reprimand jaywalkers, and guide tourists to nearby metro stations. In Hangzhou, at least 15 robots have been deployed, but they are deliberately unable to physically stop people or vehicles.
The article reports that Anthropic and Amazon have reportedly purchased large volumes of physical books, scanned them, and destroyed the originals to reduce copyright risk in AI training. It argues that this strategy is tied to differences in US and Chinese copyright law, while also raising concerns about the irreversible loss of cultural artifacts.
51World has unveiled a suite of data-collection devices and platforms designed to generate training data for embodied AI systems. The Beijing-based company argues that higher-quality physical-world data is essential for improving intelligent humanoid robots.
A new embodied AI data foundation is now available for an introductory price of 5,100 yuan. The product aims to provide full-stack physical AI infrastructure for robot training.
Wired observed a robotic arm at Generalist AI learning and improvising during a live demonstration. The robot reportedly used a banana as a tool, highlighting progress toward more adaptable embodied AI systems.
A TraceML benchmark using Qwen3-1.7B with LoRA found that equivalent effective batch sizes can have substantially different training times depending on physical batch size and accumulation steps. Across 100 optimizer updates, 4×1 was 17% faster than 1×4 on a T4 and 41% faster on an L4.
Kling AI generated more than RMB 850 million in Q2 revenue, up 240% year over year, making it the standout growth driver in Kuaishou’s otherwise slowing business. Kuaishou is prioritizing AI investment, spinning Kling AI out for independent financing at an implied valuation of US$18 billion despite near-term profit pressure.
Rapidus is pursuing mass production of advanced 2nm semiconductors through a large-scale Japanese national project. Instead of matching TSMC’s scale, the company plans to compete with RUMS, an integrated one-building production model designed for diverse, lower-volume manufacturing.
Amazon Bedrock now supports OpenAI GPT-5.6 models—Sol, Terra, and Luna—across more than 25 AWS Regions through cross-Region inference. Developers can use US geographic or global inference profiles to improve throughput and access the models through OpenAI and Converse APIs.