Robotics

To enable the interaction of a machine learning agent with its environment, a comprehensive set of elements akin to human learning - true unsupervised learning, skill acquisition, active learning, exploration, and reinforcement - are essential. However, these components remain insufficiently comprehended or effectively utilized within the prevailing supervised approaches dominating deep learning today.

Our objective is to enhance robotics through machine learning while simultaneously advancing machine learning through robotics. We actively encourage and facilitate strong partnerships between researchers in machine learning and roboticists, enabling large-scale learning on both real-world and simulated robotic systems.