Research map / Large language models / Agents, code and math
Grounding Generative Planners in Verifiable Logic: A Hybrid Architecture for Trustworthy Embodied AI
Feiyu Wu, Xu Zheng, Yue Qu, Zhuocheng Wang, Zicheng Feng, HUI LI
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace simple rejection with pedagogical causal feedback (ablation) | VIRF · SafeAgentBench | GCR: 63.4 -> 77.3 | Improved |
| replace simple rejection with pedagogical causal feedback (ablation) | VIRF · SafeAgentBench | FNR: 33.0 -> 20.2 | Improved |
| add the full verification loop to a planner given rules (ablation) | Impulsive + Rules · SafeAgentBench | HAR: 0.9 -> 0.0 | Improved |
| add RAG-synthesized knowledge to manual knowledge (ablation) | VIRF · SafeAgentBench | GCR: 70.4 -> 77.3 | Improved |
| add manual knowledge to RAG-only knowledge (ablation) | VIRF · SafeAgentBench | HAR: 11.0 -> 0.0 | Improved |
| use CoT thinker instead of impulsive planner when given full rules | Planner + Rules · SafeAgentBench | GCR: 70.5 -> 67.0 | Hurt |
| add safety rules to planner prompt | Impulsive · SafeAgentBench | HAR: 11.9 -> 0.9 | Improved |
| add diagnostic feedback to Thinker | Thinker (CoT) · SafeAgentBench | GCR: 59.1 -> 76.8 | Improved |
| add VIRF Tutor to weaker planner | Qwen-72B Thinker · SafeAgentBench | GCR: 32.0 -> 47.6 | Improved |
| use VLM-Cascade instead of hybrid detector perception (ablation) | RSSG pipeline · RSSG-THOR-30 | accuracy: 35.8 -> 76.3 | Improved |
| use VLM-Cascade instead of hybrid detector perception (ablation) | RSSG pipeline · RSSG-THOR-30 | time (s): 85.2 -> 168.4 | Hurt |
Closest papers by meaning
- SENTINEL: A Multi-Level Formal Framework for Safety Evaluation of Foundation Model-based Embodied Agents (NeurIPS 2026 · similarity 0.92)
- AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions (CVPR 2026 · similarity 0.89)
- IS-Bench: Evaluating Interactive Safety of VLM-Driven Embodied Agents in Daily Household Tasks (AAAI 2026 · similarity 0.89)
- Can We Trust Embodied Agents? Exploring Backdoor Attacks against Embodied LLM-Based Decision-Making Systems (ICLR 2025 · similarity 0.88)
- From Assumptions to Actions: Turning LLM Reasoning into Uncertainty-Aware Planning for Embodied Agents (ICLR 2026 · similarity 0.88)
- Towards Reliable Code-as-Policies: A Neuro-Symbolic Framework for Embodied Task Planning (NeurIPS 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
