人形机器人的全身控制方法、装置、电子设备及存储介质
By combining a single rigid body dynamics model and an adaptive neural network model for humanoid robots, the problem of insufficient adaptability of humanoid robots in complex environments is solved, and stable whole-body control that can quickly adapt to unknown disturbances online is achieved, thereby improving the robustness and practicality of the robot.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING SHENMOU TECH CO LTD
- Filing Date
- 2026-03-16
- Publication Date
- 2026-07-17
AI Technical Summary
In complex, unstructured environments, existing technologies make it difficult for humanoid robots to achieve real-time adaptability, leading to a decline in robustness and dynamic performance. Traditional methods rely on fixed models that fail under disturbances, while data-driven methods lack physical interpretation and generalization ability.
A single rigid body dynamic model of a humanoid robot is established. Combined with an adaptive neural network model, an initial control quantity is obtained through a model adaptive predictive controller. The model parameters are then fine-tuned online to dynamically correct the control quantity and generate joint control commands, thereby achieving whole-body control optimization.
In highly uncertain, unstructured environments, humanoid robots can quickly adapt to unknown disturbances online, achieving stable and precise full-body motion control, significantly improving robustness and practicality.
Smart Images
Figure CN121946508B_ABST