跨层级任务处理方法、电子设备、存储介质以及程序产品
By employing cross-level dynamic routing processing and a schedule uncertainty estimator, the problem of insufficient coordination between high-level semantic reasoning and low-level continuous control in embodied intelligence systems is solved, enabling efficient task processing of embodied intelligence systems in complex environments.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHEJIANG GEELY HLDG GRP CO LTD
- Filing Date
- 2026-06-22
- Publication Date
- 2026-07-17
AI Technical Summary
Embodied intelligent systems suffer from insufficient coordination between high-level semantic reasoning and low-level continuous control, resulting in poor task processing performance, especially in complex environments and under uncertain conditions where efficient decision-making and control are difficult to achieve.
By acquiring multi-layered task abstract representations of long-sequence tasks to be processed by the agent, cross-level dynamic routing processing is performed based on task progress measurement features and model confidence measurement features to generate a target execution layer representation. Based on this representation, the agent is controlled to process tasks. A progress uncertainty estimator and a hierarchical gating mechanism are introduced to achieve coordination between high-level semantic reasoning and low-level continuous control.
It significantly reduces redundant computation, improves task processing performance and response efficiency, enhances stability and flexibility in complex environments, adapts to the computing power constraints of edge agents, and reduces semantic ambiguity and decision noise.
Smart Images
Figure CN122414764A_ABST