Until Task 4 removes the duplicate WorkPlane code in grasp/visual_servo.py,
the visual-servo plane endpoints must also update perception.work_plane's
singleton so the below-plane filter (which reads from the perception side)
sees the newly-calibrated plane within the same process lifetime.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
points_via_sam21, _points_via_bbox, and _object_cloud_cv (bbox branch)
now call apply_below_plane_filter on the CV-frame cloud before the
<10-points guard. Responses carry plane_applied + n_points_before_plane.
Without a saved plane the filter is a no-op (plane_applied=false).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Owns WorkPlane (with new offset_mm field), fit_plane_svd, disk-backed
get/set, and the apply_below_plane_filter helper. WorkPlane.normal lives
in OpenCV RGB-camera frame and is oriented toward the camera. The filter
keeps points where signed_distance >= -offset_mm/1000.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extract the SAM21 wiring out of grasp_routes.py into a shared module
perception/sam21_bridge.py (Sam21Fallback + points_via_sam21). Both
/api/perception/scan and /api/perception/object_points now try SAM21
first and fall back to the bbox ROI path on any Sam21Fallback. Response
now carries 'source' (sam21 | bbox_roi) and optional 'sam21_fallback_reason'.
Side-effect: 'points_camera' on both endpoints is now in OpenCV
RGB-camera frame (same as 'points_base' minus the T_base_camera
transform), regardless of which path produced the cloud. Previously the
bbox path stored points in Nuwa-native frame here. The web GUI plots
points_base when available, so the scatter views automatically pick up
the SAM21-segmented cloud.
Verified live: /scan → source=sam21 n=2414 sam21=111ms;
/object_points → source=sam21 n=2347 sam21=118ms;
/grasp/plan regression → source=sam21 n=2390 sam21=119ms;
/scan with SAM21 down → source=bbox_roi fallback='unreachable: ...'
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Note that tests now exist and run with uv run pytest.
- Add SAM21 to the upstream services table.
- Add sam21() to the _deps factory list.
- Add a convention bullet about the SAM21 vs bbox frame asymmetry so
future contributors don't add nuwa_to_cv_frame to the SAM21 branch.
- Move 'import httpx' into the third-party imports block.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Nuwa /api/intrinsics omits width/height from the rgb and ir blocks, so
SAM21 rejected the payload with KeyError('width') and we silently fell
back to bbox on every request. Derive the dimensions from the decoded
BGR + IR depth shapes and augment intr_full before building
CameraRegistration and calling SAM21. Verified end-to-end: source=sam21,
~2200 segmented points on Mortaio at 128 ms GPU inference.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- All SAM21-path logging now lives in _points_via_sam21 at the right
level (WARNING for operational, ERROR for bug codes / unexpected).
_object_points_in_base no longer double-logs on fallback.
- Nuwa failures during SAM21 prep now raise HTTPException(502)
immediately instead of running the futile bbox fallback (which would
also fail because Nuwa is down).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When grasp.use_sam21=true (default), _object_points_in_base feeds the YOLO
bbox to SAM21 and uses the segmented cloud directly. Any failure
(transport, segmentation empty, no valid depth, payload bug) logs and falls
back to the legacy bbox path. Response meta includes source ∈ {sam21,
bbox_roi} and sam21_fallback_reason when applicable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pure refactor in preparation for the SAM21 branch. _object_points_in_base
keeps its signature and the bbox→ROI→base-frame logic moves into the
private helper. Meta now includes source='bbox_roi'.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Align sam21() with nuwa()/marker()/yolo(): require base_url to be present
in config rather than silently defaulting. Misconfiguration now raises
KeyError at factory call instead of masking as a transport error.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Catch httpx.RequestError so DecodingError etc. trigger the bbox fallback
instead of propagating raw. Lock the intrinsics dict shape in the
happy-path test, drop unused imports.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Posts (rgb_png, depth_png, intrinsics, bbox) and returns (N,3) points in
meters in OpenCV RGB-camera frame. Normalises transport vs semantic errors
into Sam21Unavailable / Sam21Error.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The 2× focal-length collision test wasn't actually colliding (adjacent IR
pixels rounded to different RGB columns). 4× makes them collapse, so the
np.minimum.at reduce is now exercised. Also drop unused H_ir, W_ir.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Forward-project IR depth pixels into the RGB pixel grid with a z-buffer.
Output uint16 mm at RGB resolution, ready to feed SAM21.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Elimina il flusso di acquisizione griglia automatica: tab GUI, modulo
autocal3d.js, AutoRunner/auto_job, e gli endpoint /auto/* + /auto_grid.
La calibrazione resta operabile via sample manuali + /solve.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
GUI:
- Rimossa tab "Calibrazione" (sostituita dalla calibrazione via AI con saved
task `calibrazione_dx`/`calibrazione_sx`, oppure dalla tab "Auto-
Calibrazione" per il flusso a griglia automatica). Backend
calibration_routes invariato — tool AI continuano a usarlo.
AI tools:
- Esposto `precise` su `move_wrist_absolute` e `move_wrist_relative`
(forwarded al campo omonimo del bridge robot, controlla la modalità di
controllo: false=veloce/tolleranza ampia, true=lento/posa accurata).
- Aggiunto `arms_go_neutral` come tool dedicato (POST /arms/go_neutral del
bridge): usato dall'AI per "torna a casa / posizione di riposo". Niente
side, agisce su entrambe le braccia.
- Fix `device_status`: ora wrappa la risposta con `_ok(...)` così il
dispatcher la classifica correttamente.
Prompt:
- System prompt aggiornato: l'AI ora passa SEMPRE `precise=true` di default
sui movimenti, escluso quando l'utente chiede esplicitamente velocità.
Sezione dedicata su come tornare a riposo (arms_go_neutral).
- Documentato che `measured_side_mm` (ritornato da get_marker_3d) viene
distorto dalla correzione affine sul depth — non usarlo per validare la
size_mm dichiarata.
Saved tasks:
- `calibrazione_dx` e `calibrazione_sx` ora partono e finiscono con
`arms_go_neutral`, cancellano i sample preesistenti per quel lato, e
acquisiscono ≥10 sample con `precise=true` su una griglia ±80mm + ±15° di
rotazione attorno alla posa home.
Config:
- Marker corrente: id=10, size=40mm (cambiato il tag fisico sul polso).
- Addendum prompt allentato: niente più "max 3 righe", ora chiede di
riportare il risultato di OGNI tool chiamato.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pipeline completa per calibrazione hand-eye e grasp planning su robot
Unitree G1/H1 con camera depth Nuwa-HP60C. Esposta come FastAPI + GUI vanilla
JS, con assistente AI (OpenRouter + tool calling curato) per pilotare il
sistema in linguaggio naturale.
Architettura:
- FastAPI orchestrator (porta da .env, default 8890) che coordina 4
microservizi upstream: Nuwa-HP60C, marker-finder, YOLO, robot bridge.
- Dual-robot: services.g1 e services.h1 in config.yaml; switchabili a
runtime via PUT /api/home/robot.
- Calibrazione hand-eye eye-to-hand split per-side (right/left).
- Marker pose: marker-finder riceve gli intrinseci RGB della Nuwa al boot,
PnP server-side con quei dati. Nessun re-PnP locale richiesto.
- Depth-to-RGB: registrazione iterativa IR↔RGB per back-projection 3D
corretta (Nuwa /api/measure usa coords IR; il nostro /api/perception/
measure_rgb prende pixel RGB e ritorna 3D nel frame RGB OpenCV).
- Correzione affine sul depth bias (calibrabile via marker come
ground-truth) — riduce errore Z da ~6% a ~1.5% nei test.
- AI assistant: 16 tool curati (move_wrist_*, hand_action, get_marker[_3d],
detect_object, calibration_*, set_active_robot…), system prompt
dinamico che riflette robot attivo + marker config + addendum
customizzabile da GUI. Temperature di default 0.1 per evitare invenzioni.
GUI (7 tab):
- Home: stato comunicazione device + selettore robot attivo G1/H1
- Assistente AI: chat con OpenRouter, history persistita lato browser,
task riutilizzabili (prompt salvati), trace tool calls
- Impostazioni AI: modello da combo curata, temperature, max iterations,
prompt addendum, anteprima system prompt
- Calibrazione, Auto-Calibrazione, TrovaUtensile, Servo Visuale, Vista 3D,
Movimento EE — invariate dalle versioni precedenti
Setup:
- Variabili in presa-mano/.env (template in .env.example)
- OPENROUTER_API_KEY richiesta solo per il tab Assistente AI
- PRESA_MANO_PORT configurabile
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
