EmbodiedWorldBench
ABot-AgentOS: A General Robotic Agent OS with Lifelong Multi-modal Memory
Jiayi Tian, Shiao Liu, Yuting Xu, Jia Lu, Zihao Guan, Honglin Han, et al. · 2026-07-17
- Modality
- Embodied & 3D
- Data
- Simulated
- Task · ground truth
- Act./Learn · Ev.+State+Act.
- Scope
- Broader evaluation with memory demands
What it tests
- CECross-modal evidenceDirectly evaluated
- LELong-range evidenceDirectly evaluated
- SCState changesDirectly evaluated
- CFConflicting memoriesNot evaluated
- UQUnanswerable queriesNot evaluated
- MAMemory-grounded actionDirectly evaluated
- SMSpatial memoryDirectly evaluated
- OSObject-state trackingDirectly evaluated
- TSTask-state trackingDirectly evaluated
Survey notes
Embodied tasks evaluate an agent with memory through navigation, object search, dialogue and replanning in changing scenes.