Multimodal Agent MemorySubmit a paper

PhotoFlow

PhotoFlow: Agentic 3D Virtual Photography Missions

Jiarui Guo, Haojia Wei, Yiming Zhang, Yifei Liu, Yuning Gong, Hongjie Zhang, et al. · 2026-05-22

Read the paperPDF

Management
T2 Dynamic Memory
Memory changes during deployment through a stable management policy.
Primary modality
Embodied & 3D
Also: text, Image
Representation
Hybrid
Distinct operative representation paths are deliberately linked or jointly consumed by downstream reasoning, generation, or control.

How experience becomes memory

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

  1. soft photographic blueprint

    1. Represented asRetained perceptual source evidenceSource- and modality-specific
    2. Stored inFlat sequential records
    3. Retrieved byMetadata / spatiotemporal retrieval
    4. Source evidenceSource directly available
    Coarse spatial cell records; region-memory summaries; forbidden zones; structured reviewer feedback; high-explore relocation controls. Director proposes new cameras from region memory, anchors, geometry probes, and high-explore relocation when needed.Full-paper method / memory representation and retrieval · PDF p. 4 · Figure 1 / method overview

  2. global anchor bank

    1. Represented asModality-specific encodings & geometric stateSource- and modality-specific
    2. Stored inFlat sequential records
    3. Retrieved byMetadata / spatiotemporal retrieval
    4. Source evidenceRetained but not accessible to the agent
    Coarse spatial cell records; region-memory summaries; forbidden zones; structured reviewer feedback; high-explore relocation controls. Director proposes new cameras from region memory, anchors, geometry probes, and high-explore relocation when needed.Full-paper method / memory representation and retrieval · PDF p. 4 · Figure 1 / method overview

  3. region memory

    1. Represented asSpatial, temporal & cross-modal structured memoryCross-represented explicit
    2. Stored inFlat sequential records
    3. Retrieved byMetadata / spatiotemporal retrieval
    4. Source evidenceSource not retained
    Coarse spatial cell records; region-memory summaries; forbidden zones; structured reviewer feedback; high-explore relocation controls. Director proposes new cameras from region memory, anchors, geometry probes, and high-explore relocation when needed.Full-paper method / memory representation and retrieval · PDF p. 4 · Figure 1 / method overview

How memory changes (T2 operations)

  • Update · Rule-based update

    During operation, PhotoFlow applies fixed-rule insertion or replacement to soft photographic blueprint; the memory state changes while the governing update policy remains fixed.

  • Revision · Evidence-weighted belief revision

    During operation, PhotoFlow applies evidence-weighted revision of stored belief or confidence to soft photographic blueprint; the memory state changes while the governing update policy remains fixed.

Related systems