Figure just live-streamed eight consecutive hours of fully autonomous, unsupervised humanoid robot operation with zero human intervention. Following the milestone, founder Brett Adcock announced the company is pushing further with an upcoming continuous 24/7 livestream to showcase around-the-clock autonomous labor.
Unitree just unveiled the GD01, a transformable manned mecha starting at $650,000. Billed as the world's first production-ready manned mecha, the civilian vehicle weighs roughly 500 kilograms with an operator inside and is capable of transforming. Unitree urged future pilots to operate the machine in a "friendly and safe manner."
Figure just showed off two F.03 humanoid robots autonomously tidying a room and making a bed together in under two minutes with zero human intervention. The robots coordinated across domestic chores in an unstructured home setup, taking another real step toward everyday household utility.
Genesis AI just unveiled GENE-26.5, a robotics foundation model paired with a human-scale hand that masters delicate physical tasks. Trained on movements captured with tactile gloves, it can crack an egg one-handed, pipette in a lab, solve a Rubik's cube, and play the piano. Why it matters: True adaptability takes robots out of rigid factories and into messy real-world tasks.
Boston Dynamics just dropped new footage of its all-electric Atlas pulling off moves that defy human biology, including an inverted handstand and a 180-degree torso rotation. Powered by a simulation-trained reinforcement learning control stack, it proves humanoids don't have to copy human joint limits to balance efficiently.
Meta just acquired humanoid robotics startup Assured Robot Intelligence (ARI) for an undisclosed sum. ARI's team will join Meta's Superintelligence Labs to build foundation models that help robots handle physical chores and learn whole-body control. Why it matters: Big tech increasingly sees physical-world interaction as the missing link to AGI, and Meta wants its AI learning hands-on.
Neuralink just revealed a custom five-axis surgical robot designed to automate brain-computer interface implants. Armed with eight cameras and real-time OCT scanners, the machine threads hair-thin electrodes directly through the dura mater without cutting it away. It paves the way for high-precision, scaled operations, though human surgeons still handle the initial craniectomy.
Sony AI just unveiled Ace, an autonomous table-tennis robot featured on the cover of Nature that defeated elite and professional human players on a full-size court under official ITTF rules. Equipped with 12 high-speed sensors yielding a 10.2 ms perception latency, Ace trained entirely in simulation to read intense spin and swing in real time. It marks a massive leap for physical AI handling split-second, high-speed physical interaction.
Honor's autonomous humanoid robot "Lightning" just won a Beijing half-marathon with a 50:26 finish, beating 12,000 human runners and topping the human world record of 57:20. Developed in a single year, the bot borrows smartphone liquid cooling and 90-95 cm legs modeled after elite sprinters. Autonomous Lightning models swept the podium, showing how fast bipedal robotics is catching up to human athletics.
Google DeepMind just launched Gemini Robotics-ER 1.6, upgrading its embodied reasoning model with sharper spatial awareness and multi-view task verification. Partnering with Boston Dynamics, DeepMind enabled robots like Spot to accurately inspect analog dials and liquid sight glasses using agentic vision. It gives industrial bots the brains to actually understand physical environments, not just blindly follow scripts.
Boston Dynamics just showcased its Spot robot tackling household chores like picking up shoes and soda cans, powered by Google's Gemini Robotics-ER 1.5 VLM. Instead of hardcoding complex state machines, engineers used conversational prompts and API tools to let the model plan and sequence actions dynamically. Why it matters: It shows how frontier visual reasoning can turn rigid industrial hardware into flexible, adaptive helpers with minimal custom logic.
Clone Robotics just revealed an ambitious roadmap to build sub-$20K musculoskeletal androids powered by fluid-driven "Myofiber" muscles. The startup aims for scalpel-wielding robot torsos by late 2026, human-like walking by 2027, and hotel robot butlers by 2028. Why it matters: If Clone can actually scale artificial muscles cheaply, it could undercut traditional electric humanoids, though its timeline still lags behind factory-deployed rivals.
Robotics startup Generalist just unveiled GEN-1, a physical foundation model that hits a 99% success rate on real-world manipulation tasks while running up to 3x faster than prior state of the art. The system is pretrained on over 500,000 hours of human wearable data and needs roughly one hour of robot-specific data to adapt to a new task. Why it matters: High failure rates have kept generalist robots off factory floors, but dependable self-recovery makes commercial deployment realistic.
Figure just brought its F.03 humanoid robot inside the White House, marking the first time a bipedal robot has visited the presidential grounds, according to founder Brett Adcock. While specific details were sparse, the symbolic milestone puts advanced robotics front and center in the highest-profile government setting.
AI giants are deploying quadrupeds like Boston Dynamics' Spot to patrol sprawling data centers around the clock. The robot dogs cost between $175,000 and $300,000 each and flag blind-spot hazards like leaks or heat anomalies. Operators claim the investment pays for itself in roughly two years by trimming physical guard headcounts.
Memories.ai just partnered with Nvidia to build a dedicated visual memory layer for smart glasses and robots. Powered by Nvidia's Cosmos-Reason 2, the system indexes first-person video so physical devices can actually recall what they see, tackling a major blind spot left by text-only AI memory.
