OmniXtreme

Key Features

• •High-DOF Articulation: The OmniXtreme humanoid robot offers a maximum of 43 degrees of freedom (EDU model), including optional waist, wrist, and dexterous hand joints. This enables complex, biomimetic movements for advanced locomotion and manipulation research. The joint ranges are extensive, with hip rotations of ±154° and knee flexion up to 165°.
• •Advanced Sensing Suite: Equipped with a depth camera and 3D LiDAR, the robot perceives its environment for navigation and obstacle avoidance. A 4-microphone array and 5W speaker provide audio interaction capabilities. The dual encoder system ensures accurate joint position feedback, critical for precise control algorithms.
• •Industrial-Grade Drive System: The joints use low-inertia high-speed internal rotor PMSM motors with industrial crossed roller bearings. Maximum knee torque reaches 120N.m on the EDU model, supporting powerful dynamic actions. Local air cooling maintains performance during extended use, and full hollow wiring protects cables.
• •Secondary Development Capability: Only the EDU model supports full secondary development, providing access to low-level joint control and sensor streams. This, combined with an optional high computing power module (e.g., Orin), allows researchers to implement custom algorithms and integrate with external systems.
• •Smart Power and Connectivity: A 9000mAh quick-release smart battery provides about 2 hours of operation. The robot supports WiFi 6 and Bluetooth 5.2 for wireless communication, and OTA updates for remote firmware upgrades, ensuring the platform stays current without physical access.
• •Modular Upgrades: Optional accessories include a three-fingered dexterous hand (Dex3-1) with force control and tactile sensors, as well as additional waist joint modules. The EDU model can be upgraded with high-computing modules, making it adaptable to varied research demands.

Use Cases

• •Locomotion Research: Researchers can program the OmniXtreme's 23 to 43 DOF to explore bipedal walking, running, and balancing. The high-torque knee joints (up to 120N.m) enable testing of dynamic gaits, leading to improvements in stability and efficiency for humanoid movement.
• •Manipulation Studies: With the optional dexterous hand and force control, the robot can perform precise grasping and object manipulation tasks. This is ideal for developing algorithms for hand-eye coordination and tactile feedback, advancing robotic dexterity.
• •SLAM and Navigation: Using the depth camera and 3D LiDAR, OmniXtreme can map and navigate unknown environments. Researchers can implement SLAM algorithms to test autonomous navigation, obstacle avoidance, and environment reconstruction in real-time.
• •Education and Training: The platform serves as a teaching tool for robotics courses. Students can learn about kinematics, dynamics, and control by programming the robot via the manual controller or secondary development kit, gaining hands-on experience with a real humanoid.
• •AI Development: The upgradeable computing module (e.g., Orin) allows running machine learning models directly on the robot. Use cases include reinforcement learning for motor skills, visual servoing, and natural language interaction through the microphone array and speaker.
• •Custom Research Platforms: The secondary development capability and modular design enable researchers to create custom experimental setups. For example, integrating additional sensors or modifying joint parameters to study specific aspects of humanoid robotics.

Who is this for?

The OmniXtreme humanoid robot is tailored for robotics researchers in academic labs and corporate R&D departments who require a high-performance platform for experimentation. It also serves advanced developers and hobbyists with a technical background in programming and mechatronics. Educators in universities can use it to demonstrate advanced robotics concepts. The EDU model specifically targets those who need full access to secondary development and customization, such as AI researchers exploring control algorithms or sensor integration. The standard G1 model suits initial exploration and validation. The product is priced for serious research applications, with the G1 at $13.5K and the EDU available upon consultation.