Human-robot collaborative motion interaction control method and device and storage medium
By constructing a causal graph of joint nodes and a human-machine coupling graph neural network model, the problem of insufficient causal relationship adaptation in existing human-machine coupling modeling is solved, and the prediction accuracy and computational efficiency of dynamic motion states are improved.
CN122401428APending Publication Date: 2026-07-17YOUDI ROBOT (WUXI) CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- YOUDI ROBOT (WUXI) CO LTD
- Filing Date
- 2026-06-05
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing human-machine coupling modeling methods rely on fixed joint association modeling, which makes it difficult to adapt to real motion causal relationships, resulting in insufficient accuracy in predicting dynamic motion states.
Method used
Construct a causal graph of the joint nodes, build a human-machine coupled graph neural network model based on the causal graph, acquire multimodal perception data and fuse motion feature vectors and semantic feature vectors to determine control parameters.
Benefits of technology
By accurately identifying the direct causal driving path of human biomechanical logic, the collaborative coupling effect of joint nodes under complex dynamic tasks is optimized, thereby improving the accuracy of state prediction.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN122401428A_ABST
Abstract
本申请公开了一种人机协同运动交互控制方法、设备及存储介质，涉及人机交互与智能机器人控制技术领域。上述方法根据关节节点的状态值，构建关节节点的因果图谱；基于因果图谱构建人机耦合图神经网络模型；获取多模态感知数据，将多模态感知数据的运动特征向量和语义特征向量融合，以获得融合后的意图特征向量；将意图特征向量作为人机耦合图神经网络模型的输入，得到人机耦合图神经网络输出的关节耦合状态序列；基于关节耦合状态序列确定控制参数，优化了复杂动态任务下各关节节点的协同耦合效果，提升了状态预测的准确性。
Need to check novelty before this filing date? Find Prior Art