Research map / Vision-language and multimodal / Robot action and navigation
End-to-end Listen, Look, Speak and Act
Siyin Wang, Wenyi Yu, Xianzhao Chen, Xiaohai Tian, Jun Zhang, Lu Lu, Yuxuan Wang, Chao Zhang
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 SA-MoE (attention-connected experts) instead of one dense model (ablation) | ELLSA · Llama Questions S2T | accuracy (%): 62.7 -> 74.7 | Improved |
| use SA-MoE instead of dense model initialized from action expert (ablation) | ELLSA · LIBERO-LONG | success rate (%): 60.6 -> 84.4 | Improved |
| SA-MoE vs dense from speech expert (unfamiliar action modality) (ablation) | ELLSA · LIBERO-SPATIAL | success rate (%): 0.2 -> 90.8 | Improved |
| integrate speech and action experts via SA-MoE vs speech expert alone | ELLSA · TriviaQA S2T | accuracy (%): 55.9 -> 45.2 | Hurt |
| integrate via SA-MoE vs action expert alone | ELLSA · LIBERO-LONG | success rate (%): 94.0 -> 84.4 | Hurt |
| speak while acting (concurrent generation) | ELLSA · LIBERO-LONG | success rate (%): 84.4 -> 73.2 | Hurt |
| speak while acting | ELLSA · TriviaQA S2T | accuracy (%): 45.2 -> 35.1 | Hurt |
| ELLSA full-duplex speech-to-speech vs Freeze-Omni | ELLSA · Llama Questions S2S | accuracy (%): 56.2 -> 70.0 | Improved |
| ELLSA vs Freeze-Omni text-mode open-ended chat | ELLSA · AlpacaEval S2T | GPTScore: 3.90 -> 3.09 | Hurt |
| ELLSA speech-conditioned VLA vs pi0-FAST | ELLSA · LIBERO average | success rate (%): 85.5 -> 89.4 | Improved |
| shorter time block (0.48 s vs 1 s) (ablation) | action expert · LIBERO | success rate | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CoT-VLA — CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models (CVPR 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- HumanSense: From Multimodal Perception to Empathetic Context-Aware Responses Through Reasoning MLLMs (AAAI 2026 · similarity 0.89)
- U-Mind: A Unified Framework for Real-Time Multimodal Interaction with Audiovisual Generation (CVPR 2026 · similarity 0.89)
- Language Model Can Listen While Speaking (AAAI 2025 · similarity 0.87)
- LifeEval: A Multimodal Benchmark for Assistive AI in Egocentric Daily Life Tasks (CVPR 2026 · similarity 0.87)
- Senses Wide Shut: A Representation-Action Gap in Omnimodal LLMs (NeurIPS 2026 · similarity 0.87)
- Talking Turns: Benchmarking Audio Foundation Models on Turn-Taking Dynamics (ICLR 2025 · similarity 0.86)
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.
