Data infrastructure firm Encord unveils a novel 300-hour multimodal dataset designed to improve robotic learning by integrating muscle-signal data with first-person video.
Encord released a new open dataset for robotic AI development, which may increase its visibility but has no direct public market impact.
Encord has launched a comprehensive open dataset featuring over 300 hours of synchronized muscle-signal and first-person video recordings. This initiative aims to enhance Physical AI development by enabling robots to interpret human hand movements beyond traditional visual data.
The dataset is designed to help robots learn from human hand movements by providing both visual and muscle-signal data, allowing for more sophisticated physical interaction.