The video shows the robot learning a towel folding task using our algorithm. The exploratory and expansion layers run 10 and 5 times, respectively. 
Once all 15 motions have been played back on the robot, the reinforcement learning iterates until convergence. Due to size restrictions, the video 
shows trials 1, 5, 7, and 10 of the exploratory layer, trials 1, 3, and 5 of the expansion layer, and trials 1 and 4 for reinforcement learning. 
The final learned motion is demonstrated at the end of the video.