🚨: A U.S. congressman claims President Donald Trump received a briefing on alleged alien-human hybrids.

Shenzhen-based Mind On has demonstrated humanoid robots and dual-arm robots working together as one team. Humanoid robots retrieve and transport items, while dual-arm robots sort and pack them at workstations. Powered by the same AI model, the robots complete the entire logistics workflow more efficiently.

Could advanced aliens already be living among us in forms we don’t recognize?

For humanoid robots to work safely in the real world, they must be able to walk, balance, recover from unexpected pushes, and handle heavy objects while adapting to changing environments. Teaching robots these skills is difficult because traditional methods often rely on either Model Predictive Control (MPC) or Reinforcement Learning (RL), each with its own limitations. Researchers have now developed a new training approach that combines MPC with RL. During training, MPC acts like a coach, providing guidance while the robot learns through trial and error. This helps the robot learn faster, more efficiently, and with greater stability than using RL alone. To make large-scale training practical, the team also created a high-speed GPU-based solver. It can train thousands of simulated humanoid robots in parallel while using less memory, significantly speeding up the learning process. The method was tested on tasks such as walking, push recovery, carrying a 10 kg payload, wearing a 10 kg vest, and pushing a 290 kg cart. It was also successfully validated on the real Themis V2 humanoid robot. Research Paper: Accelerating and Scaling MPC-Guided Reinforcement Learning for Humanoid Locomotion and Manipulation (arXiv, June 2026)

Unitree Robotics introduced Anti-Gravity mode for its G1 humanoid robot last year, improving its balance and stability. With this update, the G1 can walk, turn, perform complex movements, and quickly recover after stumbling or falling. The upgrade marked another step toward making the G1 more resilient for dynamic real-world environments.

Scientists say two stars in a distant binary system are slowly spiraling toward each other and will eventually collide, creating a rare cosmic explosion called a luminous red nova. This spectacular event will shine brighter than a normal nova but be less powerful than a supernova, releasing vast amounts of light and material into space. If the collision occurs close enough to Earth, it could even become visible to the naked eye.

Researchers have taught humanoid robots to learn skills like climbing, balancing, and handling objects on their own. Using a new framework called MotionDisco, the robots can learn without human demonstrations or remote control. The system lets robots discover different ways to complete tasks by combining AI with motion planning.

Robots can now walk blind with Figure AI's Helix Walking Controller. Instead of using cameras, the controller relies on the robot's internal sensors and reinforcement learning to maintain balance while walking. The controller, unveiled last year, allows the robot to stay balanced, recover from pushes, and move confidently across difficult terrain.

Cartwheel Robotics unveiled its Yogi humanoid robot in 2025 as a friendly AI-powered companion designed for homes. However, the company shut down in early 2026 after failing to secure the funding needed to continue development and bring the robot to market. Even with a working prototype, turning a humanoid robot into a mass-market product proved to be a major challenge.

Unitree Robotics A2 Stellar Explorer is a rugged quadruped robot designed to tackle challenging environments. It can traverse difficult terrain, carry payloads of up to 25 kg while walking, and operate for more than five hours or around 20 km on a single charge without a load. It can also climb steep slopes, making it physically capable of accompanying users on demanding hiking adventures.

1X's NEO humanoid robot is getting closer to becoming a household companion. According to the company, the first Early Access units are expected to ship to U.S. households in late 2026, with broader availability planned for 2027. Designed as a home assistant, NEO is built to help with everyday household tasks while continuously gaining new capabilities through AI and over-the-air software updates.