Multimodal Agent MemorySubmit a paper

TARIC

TARIC: Memory-Augmented Traversability-Aware Outdoor VLN under Interrupted Semantic Cues

Tianle Zeng, Hanjing Ye, Jianwei Peng, Jingwen Yu, Hanxuan Chen, Hong Zhang · 2026-05-29

Management
T2 Dynamic Memory
Memory changes during deployment through a stable management policy.
Primary modality
Embodied & 3D
Also: text, Image, GUI & computer use
Representation
Source- and modality-specific

How experience becomes memory

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

  1. World-aligned 3D cue memory particles/belief

    1. Represented asModality-specific encodings & geometric stateSource- and modality-specific
    2. Stored inFlat sequential records
    3. Retrieved byMetadata / spatiotemporal retrieval, Structure traversal
    4. Source evidenceSource not retained
    Semantic cue extraction produces 2D cue direction when visible.; Cue evidence is lifted into a world-aligned 3D memory.; Particle-like cue memory updates as the robot moves and cues appear/disappear.. spatial query over the map conditioned on current observation and instruction.Full paper: B. 3D Semantic Cue Memory · PDF p. 3 · Figure 1 / method overview

How memory changes (T2 operations)

  • Update · Rule-based update

    During operation, TARIC applies fixed-rule insertion or replacement to World-aligned 3D cue memory particles/belief; the memory state changes while the governing update policy remains fixed.

  • Revision · Evidence-weighted belief revision

    During operation, TARIC applies evidence-weighted revision of stored belief or confidence to World-aligned 3D cue memory particles/belief; the memory state changes while the governing update policy remains fixed.

Related systems