Thomas Wolf, co-founder and Chief Science Officer of Hugging Face, is extending the company’s successful open-source AI approach to robotics through the LeRobot project. Emphasizing democratization, Wolf advocates for accessible tools, diverse hardware, and community-driven development, leading to growth in Hugging Face’s robotics community to 10,000 developers. This mirrors the company’s success with transformers, showing that simple Python-based tools can empower software developers as roboticists.
Wolf champions a range of affordable robotic options, like $300 arms, over costly humanoids, promoting experimentation and broad accessibility. He stresses the importance of data diversity over sheer volume, noting that robotics requires varied datasets to generalize tasks across environments. Additionally, local model deployment enhances safety, as robots losing internet connectivity can lead to failures. Finally, Wolf believes in advancing innovation through open science, encouraging the sharing of training methodologies and datasets to foster further development in robotics, beyond merely providing pre-trained models.
Source link
