IDRAK™ supplies lightweight AR headsets for warehouse staff that overlay real‑time pick‑pack instructions while silently recording multimodal sensor data. The system streams timestamped video, gaze, and motion events into ready‑to‑use datasets for training warehouse robots, eliminating the need for scanners or separate data‑collection setups.
Funding
Funding not disclosed
Founders
Product
Problem
Warehouse fulfillment relies on handheld scanners and extensive worker training, leading to slower pick‑pack cycles and higher error rates. Additionally, manufacturers lack large, high‑quality datasets of human motions needed to train physical AI robots for automation.
Solution
IDRAK equips warehouse staff with lightweight augmented‑reality headsets that overlay step‑by‑step pick, pack, and put‑away instructions directly in the worker’s field of view, eliminating the need for separate scanners or specialized training. The glasses continuously capture video, inertial‑measurement, and gaze data, synchronizing events to 5 ms precision and automatically annotating actions such as “pick” or “carry” with object identifiers. Collected streams are anonymized by default and delivered in ready‑to‑use formats (HDF5, JSON, or custom) for robot policy learning. By turning everyday shifts into labeled datasets, the system improves current warehouse efficiency while providing the data foundation for future physical AI deployments.
Target Audience
Primary customers are fulfillment centers and distribution warehouses that need to accelerate order processing and generate high‑quality motion data for training autonomous material‑handling robots.
Features
- Real‑time AR guidance for each pick, pack, and put‑away step without requiring barcode scanners or worker retraining
- Continuous multimodal data capture (video, IMU, gaze) with timestamped cross‑modal alignment to 5 ms precision
- Automatic action labeling verified by periodic human spot checks, producing structured datasets ready for robot learning
- Data export in industry‑standard formats (HDF5, JSON) with optional custom schemas
- Built‑in privacy controls that anonymize workers and exclude personally identifiable information by default
- Integration with e‑commerce platforms (e.g., Shopify) to pull order information directly into the AR workflow