Lack of training data stifling humanoid bot development

Robotics development faces a major hurdle due to a lack of sufficient training data, especially diverse and high-quality video footage needed for humanoid robots to learn physical world interactions. Experts highlight that unlike language models trained on internet text, robots require extensive sensory and visual data to understand physics and safety in real environments.