Open Internet by MindsNet

Achieve Robot Learning from Demonstration

Develop robots that can learn complex skills by watching human demonstrations and then adapting those skills to new situations and contexts. Current robots require extensive programming or trial-and-error learning to acquire new skills. Learning from demonstration would enable robots to quickly acquire human skills through observation and imitation, then generalize those skills to new scenarios. This challenge involves advances in computer vision, motion understanding, skill representation, and transfer learning. Success would dramatically reduce the time and effort required to teach robots new tasks and enable robots to learn from humans as naturally as humans learn from each other.

Engineering, Core Engineering, Robotics