Research map / Vision-language and multimodal / Multimodal LLMs
Thinking from the Robot’s View: The CoT-HRC Benchmark for Human Intent Reasoning in Embodied Collaboration
Chenxi Deng, Chao Xu, JianmingLiu JianmingLiu, Shaofei Chen
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 |
|---|---|---|---|
| use egocentric 0.4m view instead of dual-view oracle | InternVL2.5-8B · CoT-HRC | SCAR (%): 59.6 -> 44.8 | Hurt |
| provide dual-view oracle instead of egocentric view | LLaVA-family models · CoT-HRC | SCAR (%): LLaVA-OneVision 33.4 -> 23.6; LLaVA-NeXT-Video 25.3 -> 21.7 | Hurt |
| condition on successful area and object perception (CRA vs coupled macro-intent) (ablation) | Qwen2-VL-7B-Instruct · CoT-HRC Ego_ann | accuracy (%): coupled intent 12.1-25.4 -> CRA 43.8-60.0 | Improved |
| fine-tune on 1,000 CoT-HRC instances | Qwen2-VL-7B-Instruct · CoT-HRC | SCAR (%): 34.1 -> 68.6 | Improved |
| use single final frame instead of 8 frames (ablation) | Qwen2-VL-7B-Instruct · CoT-HRC Ego | SCAR (%): 34.1 -> 41.3 | Improved |
| add generalized language instruction (help me tidy up the room) (ablation) | Qwen2-VL-7B-Instruct · CoT-HRC Ego | SCAR (%): 34.1 -> 47.3 | Improved |
| lengthen accumulative observation window from 30% to 90% (ablation) | Qwen2-VL-7B-Instruct · CoT-HRC Third view | SCAR (%): 28.0 -> 14.1 | Hurt |
| Medium vs Easy difficulty under raw ego input (ablation) | Qwen2-VL-7B-Instruct · CoT-HRC Ego | SCAR (%): Easy 34.1 vs Medium 39.4 | Mixed |
Closest papers by meaning
- BeTTER: Diagnose the Illusion of Embodied Reasoning in Vision-Language-Action Models (ECCV 2026 · similarity 0.91)
- Robobench: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models as Embodied Brain (ECCV 2026 · similarity 0.90)
- Seeing Across Views: Benchmarking Spatial Reasoning of Vision-Language Models in Robotic Scenes (ICLR 2026 · similarity 0.90)
- PARTNR: A Benchmark for Planning and Reasoning in Embodied Multi-agent Tasks (ICLR 2025 · similarity 0.89)
- EmbodiedBench: Comprehensive Benchmarking Multi-modal Large Language Models for Vision-Driven Embodied Agents (ICML 2025 · similarity 0.88)
- Seeing Together: Multi-Robot Cooperative Egocentric Spatial Reasoning with Multimodal Large Language Models (NeurIPS 2026 · 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.
