HMND 01 Alpha Bipedal

The HMND 01 Alpha Bipedal is a groundbreaking humanoid robot developed by UK-based Humanoid, unveiled on December 2, 2025. Standing at 179 cm tall and weighing 90 kg, it represents a leap in rapid development, going from initial design to a fully functional prototype in just five months—far surpassing the industry average of 18-24 months. Remarkably, it achieved stable bipedal walking a mere 48 hours after final assembly, thanks to an intensive reinforcement learning (RL) regimen in simulation using NVIDIA's Isaac Sim and Isaac Lab. This training encompassed over 52.5 million seconds of locomotion data, equivalent to nearly 19 months of traditional training, compressed into two days with minimal sim-to-real gap due to ultra-precise 3D modeling. Architecturally, the robot features 29 degrees of freedom (excluding end-effectors), balanced human-like proportions optimized for reliable bipedal motion, precision manipulation, and autonomous operation. Its locomotion capabilities are powered by an RL-trained Whole-Body Controller, enabling walking in straight and curved trajectories, turning in place, sidestepping, backwards walking, squatting, squat walking, hopping, running, and robust push recovery from omnidirectional interferences up to 350 Newtons. The modular design includes interchangeable end-effectors such as 12-DOF five-finger hands or 1-DOF parallel grippers, upper body upgrades, and protective garments, ensuring versatility and low total cost of ownership (TCO). Intelligence is driven by the VLM/VLA-based KinetIQ framework, integrating vision-language models (VLM) and vision-language-action models (VLA) for end-to-end reasoning, perception, and task execution. Powered by NVIDIA Jetson Orin AGX for GPU acceleration and Intel i9 for CPU processing, it supports low-latency, adaptive behaviors. Sensory suite includes a head with 360° vision via six RGB cameras and two depth sensors, a six-microphone array for audio, speakers, LEDs, and a head display for multimodal human-robot interaction. Body sensing incorporates 6D force/torque (F/T) sensors, wrist RGB cameras, end-effector F/T sensors, haptic feedback, and joint torque feedback, enabling precise environmental awareness and manipulation. Targeted for industrial, logistics, retail, warehouse, and emerging domestic/service environments, the HMND 01 addresses labor shortages (up to 27% in manufacturing) and unpaid care work burdens. While still in prototype stage, it builds on the wheeled Alpha variant's successful proofs-of-concept (POCs), such as bin-picking with Schaeffler. With 15 kg bimanual payload, 1.5 m/s max speed, and 3-hour swappable battery runtime, it excels in physically demanding, repetitive tasks like picking, palletizing, object handling, and human coordination. Future upgrades leverage modularity for rapid AI policy training and scalability, positioning it as a platform for real-world human-machine collaboration and quality-of-life enhancement.