Encord Uses Brain Wave Data for AI Training
Encord has started using brain wave data collected from pilots wearing specialized headsets to train more accurate and task-specific AI models. The goal is to enhance the fidelity of training datasets, which are crucial for robotics applications but notoriously difficult to scale up due to physical limitations. For developers working on machine-vision or robotic systems, this could mean a significant leap in model accuracy when dealing with complex tasks like pouring coffee or stacking objects. By incorporating brain wave data alongside traditional video and sensor inputs, Encord aims to provide a richer context for AI models to learn from. The company's San Leandro warehouse houses various items used for training robots, such as fake flowers and books. They also use leader-follower rigs to simulate tasks and collect egocentric video data. Additionally, Encord is developing forearm sensors that detect muscle signals, adding another layer of detail to their datasets.
