Multimodal Agent MemorySubmit a paper

Mobile-Agent-v2

Mobile-Agent-v2: Mobile Device Operation Assistant with Effective Navigation via Multi-Agent Collaboration

Junyang Wang, Haiyang Xu, Haitao Jia, Xi Zhang, Ming Yan, Weizhou Shen, et al. · 2024-06-03

Read the paperPDF

Management
T2 Dynamic Memory
Memory changes during deployment through a stable management policy.
Primary modality
GUI & computer use
Also: text
Representation
Cross-represented explicit

How experience becomes memory

Each path is one memory component, traced from what is stored to how it is found again.

  1. task-progress and focus-content memory

    1. Represented asSpatial, temporal & cross-modal structured memoryCross-represented explicit
    2. Stored inFlat sequential records
    3. Retrieved byMetadata / spatiotemporal retrieval
    4. Source evidenceSource linked & recoverable
    A memory unit tracks focus content and updates it with task progress for planning and decision agents.Full paper: Memory Unit · PDF p. 4 · Figure 1

How memory changes (T2 operations)

  • Update · Rule-based update

    The fixed multi-agent workflow updates focus memory as task progress changes. For Mobile-Agent-v2, this specifically supports fixed-rule insertion or replacement of task-progress and focus-content memory.

  • Revision · Rule-based state correction

    A reflection agent observes action outcomes and repairs erroneous operations. For Mobile-Agent-v2, this specifically supports fixed-rule correction of stored state of task-progress and focus-content memory.

Related systems