Catalog
Everything you need to build, repair, or upgrade humanoid robots — sourced with Canadian shipping
9 of 71 parts
DeepMind (Google)
High-fidelity physics simulator optimized for contact-rich robotics tasks. Used by virtually every humanoid robotics lab for reinforcement learning research. Apache 2.0 license.
NVIDIA
Physics-accurate robot simulation with RTX rendering, domain randomization, and sim-to-real transfer. The industry standard for training humanoid locomotion and manipulation policies.
The most powerful edge AI platform at 275 TOPS. 64GB LPDDR5, 2048 CUDA cores. Runs full transformer models, LiDAR processing, and multi-camera perception simultaneously.
Mid-range AI edge module with 100 TOPS of AI performance. 16GB LPDDR5 memory, Ampere GPU. Ideal for multi-camera perception and real-time planning on humanoid robots.
Compact AI compute module delivering 67 TOPS (Super edition). 8GB LPDDR5, Ampere GPU with 1024 CUDA cores, 2× M.2, pre-installed WiFi. The most affordable NVIDIA AI platform.
Intel
Stereo depth camera with integrated IMU providing 1280×720 depth at 90 fps. Global shutter, 87°×58° FOV, ideal range 0.3–3m. Industry standard for SLAM and robotic perception.
Open Robotics
Latest long-term-support release of the Robot Operating System. Includes MoveIt 2 for manipulation, Nav2 for navigation, and ros2_control for hardware abstraction.
Hugging Face
Open-source library for robot learning from Hugging Face. Provides pre-trained models, datasets, and pipelines for imitation learning and sim-to-real transfer.
Extended-range stereo depth camera with improved accuracy. 1280×800 depth at 90 fps, 87°×58° FOV, ideal range 0.6–6m. 2× the range and accuracy of D435 series.