ai-robotics
Coverage here spans industrial robots, cobots, humanoids, drones, and service robots, along with the hardware and software that enable them to operate at scale.
<img src="https://interestingengineering.com/_next/image?url=https%3A%2F%2Fcms.interestingengineering.com%2Fwp-content%2Fuploads%2F2026%2F11%2FUntitled-design-2026-11-12T021929.169.jpg&w=3840&q=75" alt="Google’s Gemini Decisions API processes 1,000 decisions in 150–250 milliseconds” title=”Google’s Gemini Decisions API processes 1,000 decisions in 150–250 milliseconds” loading=”lazy”>
AI and Robotics
AI and Robotics
Google’s Gemini Decisions API processes 1,000 decisions in 150–250 milliseconds
Google’s Gemini Decisions API lets developers evaluate up to 1,000 questions across text, images, audio, video and PDF inputs.
OpenAI retires ChatGPT’s two older voice modes after GPT-Live usage nearly doubles
ChatGPT is dropping two legacy voice options next month. Here’s what changes for existing users.
Google’s Gemini Robotics 2.5 scores 96% in factory trial after tackling 200 tasks
Google’s Gemini Robotics 2.5 helps humanoid robots sort automotive parts with 96% success and master more than 200 tasks.
Robots gain shared sense of touch, improved dexterity with China’s 5,000-hour dataset
LingBot-Touch unifies data from different tactile sensors, helping robots reuse touch perception across hardware platforms.
Robots learn to perform tasks faster with less human input with smart framework
Graph-as-Policy breaks tasks into reusable skills, which AI agents assemble, simulate, test, and refine into robot-control programs.
NVIDIA’s GR00T 2 transfers human skills to robots with just five task demonstrations
NVIDIA’s GR00T 2 learns physical tasks from human videos and other robots, achieving 95% success after five demonstrations per task.
Google expands its chipset lineup with up to 80% better inference price performance
Google’s new TPU 8tx promises up to twice lower latency than TPU 8t while protecting sensitive workloads.
Video: Boston Dynamics put wheels on Atlas, and now it’s pulling off 360-degree jumps
Boston Dynamics has essentially put its humanoid on wheels to see what happens.
OpenAI launches Control Room with a powerful 35-second kill switch for rogue agents
The system that can stop research workloads from generating further model outputs within 35 seconds of an intervention decision.
Muse Ember: Meta unveils its smallest model yet with 1M-token context for cheaper agents
Muse Ember handles repetitive coding and agent tasks, matching Claude Haiku 5.5 on an intelligence benchmark at nearly 3x lower cost.
Claude Fable 5.2 builds a working operating system with 1,000 agents in eight days
Anthropic’s Claude models are taking on engineering challenges that go far beyond writing code.
Defence-backed US robot builds space structure without human assistance for first time
Caltech’s ASTRA robot has autonomously assembled a truss section in orbit, marking a first step toward building large structures in space without astronauts.
Robotic legs get muscle-like flexibility with a single 3D-printed plastic
The project uses 3D-printing patterns to give a single material muscle, tendon, ligament and cushioning properties.
What comes after automation? The push to make factories more intelligent
The next industrial shift is connecting the knowledge, data, and systems that have long operated separately.
In Collaboration With
AI and Robotics
AI and Robotics
New wheeled humanoid robot learns factory jobs in a month, reports 95% success rate
AIR uses torso-mounted motors and cable-driven lightweight arms that yield on contact, enabling safer human-robot interaction.
What happens when a drone loses signal? Here’s how its failsafes work
Why losing a drone’s signal doesn’t always mean losing the aircraft.
Six humanoid robots in China work 24-hour shift in a store, two fail assessment
From making coffee to restocking shelves, humanoid robots face a retail test.
Can Claude dream now? Anthropic’s latest experiment gets eerily close to human thinking
Claude’s new dreaming approach helps it turn past experiences into organized knowledge.
Elon Musk teases SpaceX’s orbital supercomputers, says it will ‘blow people’s minds’
The Tesla and SpaceX chief also revealed Starlink’s next-generation capabilities and Tesla’s chipmaking ambitions.
1,600x faster: Google DeepMind agents expose weaknesses in older encryption methods
The company said its agents, guided by human supervision, rapidly solved mathematical problems that once required extensive manual research.
About AI and Robotics
AI and robotics are changing how machines behave in the real world. Not in theory or demos, but on factory floors, in warehouses, on roads, and increasingly in everyday spaces. This category focuses on how intelligence is being built into machines, what actually works, and where the gaps still are.
At one end are physical systems: robots that move, grip, balance, see, and sense their surroundings. At the other end is software: machine learning models, perception stacks, control systems, and decision-making logic that determine how those machines respond when the environment doesn’t behave as expected. Most of the interesting work happens in the messy middle, where sensors misread, data is incomplete, and systems have to operate safely around people.
AI and robotics coverage at Interesting Engineering spans industrial robots, cobots, humanoids, drones, and service robots, along with the hardware and software that enable them to operate at scale. That includes sensors, actuators, batteries, AI chips, simulation tools, and the increasingly important role of edge computing. We also look at how these systems are trained, tested, deployed, and maintained after leaving the lab.
This category examines constraints such as power, cost, reliability, latency, safety, and regulation, and why many promising systems struggle outside controlled environments. Human–machine interaction, deployment failures, workforce impact, and real-world trade-offs are part of the story.
As robots move out of fenced-off spaces and into shared ones, the question isn’t just what they can do, but whether they can be trusted to do it consistently. AI and Robotics tracks progress where it’s real, skepticism where it’s warranted, and the engineering work still needed to turn intelligent machines from impressive prototypes into dependable tools.
It also looks at who is building these systems and why the gap between research success and commercial reliability remains one of the field’s hardest problems.
