# 论文草稿：骨架

> 本页由仓库源文件生成。请不要直接编辑本页；修改下方 source 文件后运行 `vp run docs:generate`。

> 中文支持说明：本页是中文站点镜像，保留论文标题、证据状态和来源字段的原始表述，避免未经核实的翻译改变证据边界。

> 来源: `paper/manuscript.md`

## SAH-BRI-Grasp Manuscript Skeleton

Status: `evidence-gated skeleton ready`

Do not write strong scientific conclusions here until evidence rows and paper cards support them.

## Working Title

SAH-BRI-Grasp: A Scene-Aware Hybrid Brain-Robot Interface for Robotic Grasping

## Framework Boundary

* Framework: `SAH-BRI` - Scene-Aware Hybrid Brain-Robot Interface.
* Current system: `SAH-BRI-Grasp` - Scene-Aware Hybrid Brain-Robot Interface for Robotic Grasping.
* Current evidence corpus: 73 local PDFs, 73 extracted text artifacts, and 73 structured paper cards (`72 extracted` + `1 extracted from OCR`).

## Current Evidence Boundary

| Claim Area | Current Status | Writing Rule |
| --- | --- | --- |
| Project scope | `verified` | Can state as the paper's target system and research direction. |
| Shared-control formalism | `verified` | Can cite `SA-001` and `SA-002` as baseline formalisms. |
| EEG as high-level intent | `inferred` | Can frame as evidence-motivated design rationale, not as final performance conclusion. |
| SSVEP scene-target role | `inferred` | Can justify with SSVEP frequency-tagging/decoder evidence, but dynamic object-box performance remains experimental. |
| MI mode-control role | `inferred` | Can justify as a design choice from low-dimensional MI robotic control evidence. |
| YOLO dynamic command space | `inferred` | Can describe as system contribution hypothesis, not proven novelty. |
| Robot grasping boundary | `inferred` | Can state YOLO is not grasp execution; grasp-pose and project calibration implementation remain incomplete. |
| Results and user workload | `needs confirmation` | Do not write result claims before experiments. |

## Planned Structure

| Section | File | Current Action |
| --- | --- | --- |
| 1. Introduction | `paper/sections/01-introduction.md` | Build claim-gated scaffold. |
| 2. Related Work | `paper/sections/02-related-work.md` | Draft ready from the active 73-paper corpus; keep system-level claims as design rationale. |
| 3. Method | `paper/sections/03-method.md` | Interface skeleton expanded from system pipeline, runtime records, state machine, and safety design. |
| 4. Experiments | `paper/sections/04-experiments.md` | Expand protocols and baseline definitions before data collection. |
| 5. Results | `paper/sections/05-results.md` | Placeholder skeleton only; keep blocked until experiment outputs exist. |
| 6. Discussion | `paper/sections/06-discussion.md` | Limitation scaffold drafted; fill result interpretation after experiments. |
| 7. Conclusion | `paper/sections/07-conclusion.md` | Placeholder skeleton only; keep blocked until results and discussion are evidence-backed. |

## Docs Linkage

The Vocs Paper Kit exposes the manuscript overview, skeleton next steps, and all seven draft sections. Generated docs must be refreshed with `vp run docs:generate` after source edits.

## Current Writing Boundary

The manuscript may now contain:

* framework/system definitions;
* evidence-backed motivation and design rationale;
* method and experiment skeletons;
* explicit gaps and limitations.

The manuscript must not yet contain:

* claimed closed-loop task success;
* claimed grasp success;
* claimed user workload reduction;
* claimed low-channel performance;
* project-specific hand-eye calibration procedure or accuracy claims, because local hardware calibration is not yet defined or validated.