From the archive
AI robotics combines perception, planning and learning with machines that act in the physical world. Follow developments in humanoid robots, autonomous systems, robotic manipulation and embodied intelligence. Our reports examine what was demonstrated, the conditions it worked under and the gap between a controlled trial and reliable deployment, with sources you can check and space to discuss the results.
Lead story
Google will host more than 100 AI startups at a two-day summit in Mountain View this November, with founders set to work with Google DeepMind and Google Cloud specialists. The programme includes hands-on technical support and up to $350,000 in Google Cloud credits for eligible startups, making this…
Read the report & discussion ↗Watch Desk
South Korea has selected Doosan Robotics to lead two government-backed projects with a combined research budget of about $74 million, spanning AI-enabled collaborative robots and automated welding. The plans put locally processed AI into industrial machines, with commercial robot products targeted for 2031.
NVIDIA Watch
Nvidia and Hon Hai, better known as Foxconn, say robots have improved assembly of components for Nvidia’s GB300 AI computing platform. The reported success rates top 90 per cent, offering a concrete look at where flexible automation is being put to work in AI hardware manufacturing.
Watch Desk
Reka AI has introduced Rho-1, a 19-billion-parameter model built to handle text, images, video and robot-control actions in one network. The notable idea is a shared model and context for tasks that are often split across specialised systems, including work involving robots.
Watch Desk
Extend Robotics has raised £2.6 million to expand a service that lets manufacturers pay for completed robotic work rather than buy robots outright. The UK company says the funding will also support a remote-operator network, an approach that could put robotic capacity within reach of factories wary of…
Watch Desk
Hitachi and Germany’s Agile Robots are partnering to develop physical AI systems for complex factory tasks, combining Hitachi’s edge AI technology with Agile’s robotics platform. The plan is to commercialise the resulting systems globally through Hitachi’s HMAX Industry portfolio, putting AI closer to machines and production lines rather…
Google DeepMind Watch
Google DeepMind has released Gemini Robotics 2, a robotics model suite combining an embodied reasoning model with on-device controllers. Research lead Keerthana Gopalakrishnan says the work also tackles whole-body control, while the field still has a long way to go on getting systems to generalise across different robot…
AMD Watch
AMD has reportedly agreed to buy World Labs, the spatial-AI startup co-founded by Fei-Fei Li, in an all-stock deal valued at about $8.2bn. The reported move would bring a team working on AI models of three-dimensional worlds into a chipmaker’s effort to anticipate what computing comes next.
Watch Desk
Business Insider reports that robotics start-up FieldAI has raised $700 million at a $10 billion valuation, up from $2 billion in 2025. The reported round is a striking bet on a company building a general-purpose AI system for robots, from humanoids to drones.
Watch Desk
Researchers say a code-execution approach helped robot agents complete more simulated tasks while using fewer language-model resources. In 700 task instances, PyRUA-Lean raised reported success from 63.1% to 71.7% under equal LLM-call budgets, a result that could matter for building more efficient robot agents.
MIT CSAIL Watch
MIT CSAIL researchers and collaborators have developed InstructMesh, a tool that lets people refine AI-generated 3D models before fabrication. In a reported study, novices identified and fixed design flaws about 90 per cent of the time, even though nearly 80 per cent of the models tested had structural…
Watch Desk
Developers told VentureBeat that safeguards in OpenAI and Anthropic models are flagging ordinary aerospace, robotics and security tasks, interrupting work and sometimes pushing them towards other models. The dispute is practical: how can AI tools block dangerous requests without treating routine engineering as a threat?
