Scalable Representations for Vision and Robotics