robomemory-multi-memory-embodied-ai - RoboMemory Multi-Memory Framework for Embodied Agents
Integrates spatial, temporal, episodic, and semantic memory for embodied agents with dynamic spatial graphs, closed-loop planning, and critic feedback.
Tags
Updated: 2026-10-15Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Maintain spatial knowledge graphs
- Track temporal events
- Store episodic experiences
- Build semantic knowledge
- Generate long-horizon plans
- Evaluate plan feasibility
- Provide adaptive feedback
Inputs
- Environment observations
- Object positions and types
- Observation timestamps
- Action parameters
- Action outcomes
- Action preconditions
- Task trajectories
Outputs
- Updated spatial maps
- Temporal event records
- Episodic experience records
- Semantic knowledge facts
- Long-horizon plans
- Plan feasibility evaluations
- Adaptive planning feedback
- Spatial context queries
- Graph paths
Requirements
- Python environment
- PyTorch
- NumPy