A competition-proven, fully bipedal humanoid built for education and research. With 22 degrees of freedom, NVIDIA-powered AI, and ROS 2 compatibility, the K-1 brings advanced robotics into your classroom.
The K-1 combines competition-grade bipedal locomotion with an open development platform, purpose-built for STEM education and AI research.
Powered by NVIDIA Jetson Orin NX delivering up to 200 TOPS of AI inference. Run neural networks, computer vision, and behavior control entirely on-board without cloud dependency.
22 degrees of freedom with full-force-controlled joints and dual encoders enable natural walking, balancing, kicking, and dynamic athletic movements — validated at RoboCup 2025.
Fully compatible with ROS 2 and open APIs. Program in Python, C++, or use simulation environments like Isaac Sim, MuJoCo, and Webots for training and testing.
3D depth camera for spatial perception, 9-axis IMU for balance, microphone array for voice interaction, speaker for audio output, and touch sensors for safe human interaction.
At just 95 cm tall and 19.5 kg, K-1 fits in a carry case. Move it between classrooms, labs, and competitions with ease. Battle-tested durability from RoboCup championships.
Integrated simulation and reinforcement learning environment. Train locomotion policies, develop AI behaviors, and deploy directly to hardware — bridging sim-to-real robotics.
From introductory STEM courses to PhD-level AI research, the K-1 provides a real-world platform for the full spectrum of robotics education.
Students program real bipedal walking, obstacle avoidance, and object recognition. The K-1 transforms abstract programming concepts into visible, physical outcomes that inspire deeper learning.
K–12 & Higher EdDeploy reinforcement learning policies, computer vision models, and NLP pipelines on real hardware. The NVIDIA computing platform handles YOLO, pose estimation, TTS, and ASR on-board.
University LabsThe RoboCup 2025 KidSize champion platform. Both the 1st and 2nd place teams competed on identical K-1 hardware — making it a pure test of algorithms and software innovation.
RoboCup & CompetitionsStudy embodied AI, social robotics, and natural language interaction. The K-1's multimodal sensor suite — vision, audio, touch — provides a complete platform for HRI research.
ResearchFull-force-controlled joints, dual encoders, and NVIDIA AI computing in a portable 19.5 kg frame. Every spec optimized for education and research.
Download Full Spec SheetFrom visual programming to advanced sim-to-real reinforcement learning — the K-1 supports every level of development.
Full ROS 2 support with open APIs for joints, sensors, and high-level motion interfaces. Integrate with the broader robotics ecosystem.
Integrated simulation environment for training locomotion and AI behaviors. Compatible with Isaac Sim, MuJoCo, and Webots for sim-to-real transfer.
Configure networking, remote control, and run animations from the companion mobile app. Supports LLM integration with TTS and ASR for natural language interaction.
Speak with a uLearn specialist about the K-1 Humanoid. Available in Geek, Edu, and Pro configurations to match your needs and budget.