Ateegnas
01 / EMBODIED INTELLIGENCE & ROBOTICS

Capture how thephysical world works

Household and everyday first-person POV video collection designed for generalist robotics, embodied AI, and physical world foundation models

4,500+ unscripted hours · 100% de-identified
02 / SHG DATA ARCHITECTURECOMMUNITY-DRIVEN CAPTURE
01Bimanual POV

SHG Grassroots Network

Distributed network of trained Self-Help Group (SHG) women across rural, semi-urban, and urban India capturing authentic everyday physical manipulation tasks.

02Spatial Affordance

Head-Mounted Capture Rig

Standardized ultra-lightweight 4K 60fps headgear aligned precisely with human eye-level gaze for true first-person ego-pose and hand-object interactions.

03100% De-Identified

Automated Privacy Face-Blur

Every captured hour undergoes edge-based face blur, bystander redaction, and audio strip before ingestion into our multi-agent labeling pipeline.

04Robotics Ready

Embodied Action Indexing

Micro-action temporal segmentation (e.g. reach, grasp, stir, fold) paired with spatial affordance masks ready for humanoid training.

03 / DOMESTIC TASK DOMAINSBIMANUAL MANIPULATION

Kitchen & Meal Preparation

1,800+ Hrs

Complex fine-grained hand-tool manipulation with varied textures, temperatures, and utensil types.

Dough kneadingVegetable choppingSpice grindingStove ignitionVessel washing

Domestic Cleaning & Organizing

1,100+ Hrs

Long-horizon gross and fine motor coordination in cluttered 3D domestic room geometries.

Broom sweepingWet moppingDust wipingAppliance operationTrash classification

Textile & Garment Care

900+ Hrs

Deformable object manipulation requiring dynamic tactile adjustment and shape estimation.

Fabric sortingBucket washingSun-line hangingPrecision garment foldingStorage stacking

Elderly & Daily Living Care

700+ Hrs

High-safety human-adjacent assistive routines with compliant velocity and force constraints.

Medicine sortingTray carryingWalking assistanceDoor latchingBed making
04 / CUSTOM SHG DATASET INQUIRY

Configure Your Physical AI Data Requirements

100 Hrs (Pilot)1,000 Hrs5,000+ Hrs (Foundation)

All footage is processed through automated on-device face blur, de-identification, and audio redaction.