COBALT: Crowdsourcing Robot Learning via Cloud-Based Teleoperation with Smartphones
Researchers have developed COBALT, a cloud-based teleoperation platform that uses smartphones to collect large-scale robot learning data. This system allows multiple users to contribute to robot training simultaneously, significantly reducing costs and enabling data collection from almost anywhere. A user study confirmed that smartphone-based operation is as effective as specialized hardware, leading to the collection of over 7500 demonstrations across nine countries in just five days. AI
IMPACT Enables scalable and cost-effective data collection for robot learning, potentially accelerating advancements in the field.