This commit is contained in:
NgoQuocViet2001 2026-06-04 05:15:19 +08:00 committed by GitHub
commit 65235ba739
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 29 additions and 2 deletions

View File

@ -2,11 +2,11 @@
| Field | Value |
|-------|-------|
| **Status** | Proposed |
| **Status** | Proposed + amended by ADR-113 |
| **Date** | 2026-03-02 |
| **Deciders** | ruv |
| **Codename** | **RuvSense** -- RuVector-Enhanced Sensing for Multistatic Fidelity |
| **Relates to** | ADR-012 (ESP32 Mesh), ADR-014 (SOTA Signal Processing), ADR-016 (RuVector Training), ADR-017 (RuVector Signal+MAT), ADR-018 (ESP32 Implementation), ADR-024 (AETHER Embeddings), ADR-026 (Survivor Track Lifecycle), ADR-027 (MERIDIAN Generalization) |
| **Relates to** | ADR-012 (ESP32 Mesh), ADR-014 (SOTA Signal Processing), ADR-016 (RuVector Training), ADR-017 (RuVector Signal+MAT), ADR-018 (ESP32 Implementation), ADR-024 (AETHER Embeddings), ADR-026 (Survivor Track Lifecycle), ADR-027 (MERIDIAN Generalization), ADR-113 (Multistatic Anchor Placement Strategy) |
---
@ -42,6 +42,12 @@ Quantified:
- 20 Hz output rate (50 ms cycle time)
- Breathing SNR > 10dB at 3m (validates small-motion sensitivity)
### 1.4 Placement Amendment (ADR-113)
ADR-113 amends this ADR with the installation guidance that was intentionally left open here: how many anchors to deploy, where to place them, and how that decision changes by cog category and target zone. Implementers should treat ADR-113's 4-axis placement matrix as the binding anchor-placement specification for RuvSense deployments.
The practical consequence is that "4 nodes" is no longer a universal default. The recommended anchor count is cog-dependent: presence can start at 3 anchors, person-count at 4, 2D pose/vital-signs at 5, and 3D vital-signs at 6 low/mid wall anchors with no ceiling-only layout.
---
## 2. Decision
@ -129,6 +135,26 @@ Synchronization: GPIO pulse from aggregator node at cycle start. Clock drift at
2. `DynamicMinCut` partitions into K clusters (one per detected person)
3. Attention fusion (§5.3 of research doc) runs independently per cluster
#### 2.4.1 Anchor Placement Matrix (ADR-113 Amendment)
ADR-113 fills ADR-029's open placement question with a cog-aware decision matrix:
| Cog category | Dimension | Zone mode | Occupants | Recommended anchors | Anchor heights | Expected coverage |
|--------------|-----------|-----------|-----------|---------------------|----------------|-------------------|
| Presence / occupancy | 2D | body | 1 | 3 | walls @ 0.8 m | 63% |
| Person count | 2D | body | 1-4 | 4 | walls @ 0.8-1.5 m mixed | 86% |
| Pose estimation | 2D | body | 1-2 | 5 | walls @ 0.8/1.5 m mixed | 97% |
| Vital signs | 2D | chest | 1-4 | 5 | walls @ 0.8/1.5 m | 100% |
| Pose estimation (3D) | 3D | body | 1-2 | 7-8 | mixed: 0.8/1.5/2.4 m | 65%+ |
| Vital signs (3D) | 3D | chest | 1-4 | 6 | walls @ 0.8/1.5 m, no ceiling | 82% |
Placement rules inherited from ADR-113:
1. Avoid ceiling-only deployments; they place the Fresnel envelope above floor- and chest-level targets.
2. Match anchor heights to target zones: chest-centric vital-sign cogs prefer low/mid wall anchors, while full-body 3D pose can use mixed heights.
3. Keep multi-feature cogs at N >= 4 so R7 mincut single-anchor-compromise detection remains available.
4. Use the future `wifi-densepose plan-antennas --cog <name>` CLI path to select the matrix row and produce a deterministic installation diagram.
### 2.5 Coherence Metric
Per-link coherence quantifies consistency with recent history:
@ -387,6 +413,7 @@ No new workspace dependencies. All ruvector crates are already in the workspace
| ADR-024 | **Used**: AETHER embeddings for person re-identification |
| ADR-026 | **Reused**: Kalman + lifecycle infrastructure lifted to RuvSense |
| ADR-027 | **Used**: GeometryEncoder, HardwareNormalizer, FiLM conditioning |
| ADR-113 | **Amends**: Adds the cog-aware anchor-count and placement matrix for RuvSense deployments |
---