# SAH-BRI

> Generated from repository source files. Do not edit this page directly; edit the source file listed below and rerun `vp run docs:generate`.

> Source: `framework/sah-bri.md`

## SAH-BRI

Scene-Aware Hybrid Brain-Robot Interface

## Framework Definition

`SAH-BRI` is the repository-level framework for scene-aware hybrid brain-robot interaction. It combines visual scene understanding, non-invasive hybrid BCI channels, robot shared autonomy, and runtime safety gates to convert high-level human intent into robot action.

Claim status: `verified` as a project framework definition.

## Current System Instance

`SAH-BRI-Grasp` is the current system instance: Scene-Aware Hybrid Brain-Robot Interface for robotic grasping.

The working pipeline is:

```text
visual scene understanding
        -> selectable object candidates
        -> SSVEP target selection
        -> MI mode control or intervention
        -> shared autonomy and safety arbitration
        -> robot planning and grasp execution
        -> verification and recovery
```

Claim status: `verified` as the current system scope; project-specific performance remains `needs confirmation`.

## System Extension Boundary

The SAH-BRI framework can support more than one system instance without changing the current research scope:

```text
SAH-BRI
|-- SAH-BRI-Grasp     current robotic grasping system
`-- SAH-BRI-Manip     provisional future dexterous-manipulation name
```

Mounting a dexterous hand does not by itself rename `SAH-BRI-Grasp`. If the hand executes a small set of grasp primitives while the primary experiment remains scene-aware target selection and grasp completion, it is an end-effector option inside the current system.

The theme changes only when task intention, skill selection, contact adaptation, or dexterous manipulation becomes a primary experimental variable. At that point, `SAH-BRI-Manip` may become a separate system instance and paper direction after its hardware, evidence, protocol, and baselines exist.

Status: `verified` as a project scope decision; `SAH-BRI-Manip` remains a provisional name and its feasibility is `needs confirmation`.

See [Dexterous Arm-Hand Extension](/research-map/dexterous-arm-hand-extension) for the staged route and promotion gates.

## Design Rationale

The framework is based on a constrained interpretation of non-invasive EEG:

* SSVEP is treated as a discrete target-selection and confirmation channel.
* MI is treated as a low-bandwidth mode-control, stop, or intervention channel.
* Scene perception generates the command space instead of forcing the user into fixed low-level directional commands.
* Shared autonomy maps uncertain high-level intent into feasible robot action.

Claim status: `inferred` from component literature and repository evidence. The integrated closed-loop system still requires local experiments.

## Evidence Boundary

The 73-paper local corpus supports component-level rationale for BCI foundations, SSVEP, MI, object detection, grasping/calibration, hybrid BRI, and shared autonomy. It does not yet support final claims about:

* closed-loop grasp success;
* user workload reduction;
* reliable low-channel online operation;
* physical hand-eye calibration accuracy;
* clinical or rehabilitation efficacy.

Those claims must remain `needs confirmation` until experiment records and evidence rows are available.

## Next Step

Continue to the [SAH-BRI-Grasp System Overview](/framework/sah-bri-grasp).