It continuously calculates the Zero Moment Point (ZMP) at an extremely high frequency or utilizes Whole-Body Control (WBC) to adjust the angles of each leg joint in real-time. This enables stable walking, running, and jumping, handling emergencies like fall protection and autonomous recovery, all while coping with uneven ground, external pushes, or even slipping. Current typical industry solutions employ ARM or X86 CPU architectures.
Compared to wheeled robots, these robots have more stringent requirements for self-posture control and spatial recognition, constantly facing challenges to kinematic stability during operation. They possess higher sensor integration and processing capabilities, enabling them to execute long-sequence task planning (e.g., "navigate through multiple rooms to find and close a valve") and perform 3D reconstruction and navigation in complex environments. Current typical industry solutions involve high-performance GPUs like Orin development kits, but these kits often face issues with insufficient sensor integration capabilities and bandwidth limitations.
Cerebellum: T80 utilizes high-performance ARM or N204 high-performance X86.
Brain: N201/T300/T500 supports a high-performance "brain" for multi-sensor integration and high computational power.