25. Use deep learning for robot perception
Apply neural networks to perception tasks such as detection, segmentation, depth, grasp prediction, and scene labeling. Cover datasets, annotation, training, evaluation, edge inference, and failure modes that matter on robots.