Research projects
STAIR (STanford AI Robot) project
Integrating tools from all the diverse areas of AI, to build a useful, general purpose home assistant robot.
Building 3d models from a single still image. (Online demo available.)
Machine learning for high-precision aerobatic helicopter flight.
Learning algorithms to enable a four-legged robot to climb over obstacles and negotiate rugged terrain.
Automatically enlarging WordNet, using machine learning.
Other reinforcement learning videos
High-speed obstacle avoidance, snake robot, etc.
Self-taught learning and Neuroscience-informed learning
Algorithms for learning from unlabeled data, and learning algorithms informed by neuroscience computational principles.