Intro

Lifelong Embodied Agents

LifecycleA lifelong agent keeps learning, reusing knowledge, and becoming more efficient.

Learning01

Locality

Cropped IsCiL Figure 1 showing only prototype-based skill incremental learning.

What should be updated?

Utilization02

Compatibility

Top section of SIL-C Figure 1 illustrating policy-compatible skill incremental learning.

Will new knowledge remain compatible?

Acceleration03

Cost Efficiency

RECPO Pareto graph comparing ordered completion and task inference time across skill-context diversity.

Can the agent do more with less?