Multimodal Agent MemorySubmit a paper

MobileUse

MobileUse: A GUI Agent with Hierarchical Reflection for Autonomous Mobile Operation

Ning Li, Xiangmou Qu, Jiamu Zhou, Jun Wang, Muning Wen, Kounianhua Du, et al. · 2025-07-21

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. proactively explored app knowledge

    1. Represented asSpatial, temporal & cross-modal structured memoryCross-represented explicit
    2. Stored inHierarchical multi-tier store
    3. Retrieved byPlanner / tool-assisted retrieval
    4. Source evidenceSource linked & recoverable
    Self-planned exploration builds reusable knowledge of unfamiliar mobile environments.Full-paper method / memory representation and retrieval · PDF p. 2 · Figure 1

  2. hierarchical reflection history

    1. Represented asPerception-to-text records & summariesCross-represented explicit
    2. Stored inFlat sequential records
    3. Retrieved byMetadata / spatiotemporal retrieval
    4. Source evidenceSource linked & recoverable
    Action-, trajectory-, and global-level reflection retain task progress and error evidence.Full-paper method / memory representation and retrieval · PDF p. 6 · Figure 1

How memory changes (T2 operations)

  • Consolidation · Experience abstraction

    Proactive exploration converts interactions into reusable app knowledge. For MobileUse, this specifically supports abstraction of episodes into reusable experience of proactively explored app knowledge.

  • Revision · Rule-based state correction

    Hierarchical reflectors detect and recover from errors under fixed triggers. For MobileUse, this specifically supports fixed-rule correction of stored state of proactively explored app knowledge.

  • Update · Rule-based update

    During operation, MobileUse applies fixed-rule insertion or replacement to proactively explored app knowledge; the memory state changes while the governing update policy remains fixed.

Related systems