Nexdata Delivers 10,000 Hours of Synchronized Ego-Exo Data for Robot Learning

Nexdata, a provider of off-the-shelf datasets, Bespoke data collection, and data annotation services for AI development, has delivered a significant dataset for Physical AI research. The company has collected 10,000 hours of synchronized ego-exo data, capturing hand-object interactions and full-body movements across various factory and warehouse environments. Each data collection setup involved five synchronized devices: one stereo head-mounted device, two wrist-mounted cameras, and two external cameras. These devices recorded interactions, workspace context, and full-body movements, providing researchers with a comprehensive view of the actions being performed. The data was collected in eight-hour sessions to ensure environmental diversity, covering tasks such as component assembly, tool handling, goods sorting, shelf picking, and packing.

Nexdata employed a rigorous quality control process, combining AI-assisted checks with manual review to ensure the authenticity of tasks, camera synchronization, visibility, and recording completeness. Reviewers also checked the consistency of annotations and verified that corresponding files from every view were present before delivery. The project achieved a data quality pass rate of up to 98%, demonstrating Nexdata’s capability to coordinate multi-device collection, varied work environments, and quality control at scale for robot learning projects.

The delivery of this dataset is a testament to Nexdata’s expertise in supporting Physical AI development through Bespoke data collection, annotation, and integrated data services. The company’s Physical AI Data Solution, which includes off-the-shelf datasets for training and evaluation, is now available for researchers and developers. This dataset can be explored in the full case study available at www.nexdata.ai/company/news/1478?fromPage=1&fromTag=.

Beyond individual projects, Nexdata supports Physical AI development through Bespoke data collection, annotation, and integrated data services, alongside off-the-shelf datasets for training and evaluation. The company’s Physical AI offering combines collection hardware, data management and annotation platforms, and professional services to support clients’ data pipelines. For more information on this project and Nexdata’s offerings, readers can visit the company’s website at Nexdata.