Design per: detection tag su tutte le camere attive (Nuwa rosso + D435i),
rinomina "Tag camera frontale" -> "Tags", calibrazione persistente
Nuwa->frontale via tag condiviso (Umeyama similarity), flag+pulsante
"Allinea" nella sezione Cloud cross-camera.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Rimosso il modulo agent (LLM chat) e la relativa config OpenRouter.
La tab Pose 3D ora: espone i giunti reali via G1 /diagnostics per
specchiare l'URDF senza IK; acquisisce point cloud Nuwa+D435i
(/api/pose_3d/clouds); mostra il modello 3D della camera sul polso;
degrada gracefully a 200 (wrist=null) quando il robot è offline.
Il vecchio renderer canvas-2D (app.js view3d + diagnostic3d.js) è
stato consolidato nel viewport Three.js di pose_3d.js.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Plan dettagliato per la tab "Pose 3D" (writing-plans skill),
basato sullo spec del 2026-05-21. 15 task suddivisi in backend
(routes + tests TDD) e frontend (Three.js viewport + gizmo +
mode toggle Editor/Jog).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Specifica per nuova tab GUI con viewport Three.js, gizmo
translate/rotate sui polsi DX e SX, modalita' Editor (preview) /
Jog (auto-move al mouseup), safety guard sui movimenti grandi,
checkbox precise toggle, save/load named poses in config.yaml.
Brainstorming nella sessione 2026-05-21.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
13 task TDD per il nuovo tab: config, geometry, math puro, 6 endpoint
router /api/right_arm/*, markup HTML + CSS, JS module. Spec:
docs/superpowers/specs/2026-05-20-right-arm-tab-design.md
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Design del nuovo tab "Braccio DX": fronte (depth cam fissa sul robot)
+ polso (Nuwa eye-in-hand) + controlli move/home/custom-pose. Router
/api/right_arm/*, JS dedicato, persistenza in config tab_right_arm.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Brainstorming → approved spec for the GUI/backend redesign of tab "Cerca
modello": YOLO+SAM21 cluster feeds Open3DModel /api/pose, presa-mano
samples the recipe STL, projects onto a snapshot RGB with togglable
overlays for accepted/excluded points and the fitted model points.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
6 task TDD-first: perception/work_plane.py + 8 unit tests, filter wiring
ai 3 call sites con plane_applied/n_points_before_plane nelle response,
nuovi endpoint /api/work_plane/{fit,save,get,delete}, rimozione totale
del visual-servo (backend + GUI), nuova tab Calibrazione Piano con
work_plane.js (ROI + fit preview + 3D viewer con scatter colorato +
offset live-recolor), verifica live end-to-end.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Repurpose the "Servo Visuale" tab into "Calibrazione Piano". Reuses the
existing WorkPlane + fit_plane_svd infrastructure (moved out of
grasp/visual_servo.py into perception/work_plane.py), extends WorkPlane
with offset_mm, splits fit (preview) from save (commit), and applies a
below-plane + offset filter to every object cloud (sam21_bridge,
_points_via_bbox, _object_cloud_cv bbox branch). Visual-servo feature
removed entirely; the only survivors are the plane bits.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Spec brainstorming validato: SAM21 si inserisce tra YOLO bbox e PCA con
fallback automatico al path bbox legacy. Tre componenti nuovi
(clients/sam21.py, perception/depth_align.py, helper _points_via_sam21)
e due modifiche minori (config.yaml, api/_deps.py). Out-of-scope: scan/
object_points, mask UI, nuwa_to_cv_frame cleanup.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
