- Posted September 22, 2013 by
Albuquerque, New Mexico
This iReport is part of an assignment:
Quadrotor delivers coffee
One such task is a coffee delivery. Coffee delivery is an interesting application because it has to be done while the coffee is hot (time-sensitive), without spilling it (swing-free), and without hitting obstacles (collision-free).
Researchers at University of New Mexico, use reinforcement learning to teach UAVs to deliver coffee. Reinforcement learning rewards desired behavior - when the UAV delivers the drink without spilling it, and punishes undesired behavior. Check out the movie: https://www.youtube.com/user/sandraorionnm